JPH07152787A - Information access system and recording medium - Google Patents

Abstract

PURPOSE:To improve the operability of a system and to set the system to be convenient even for a physically handicapped user by permitting a center system to transmit phonetic symbol information to a terminal and the terminal to receive phonetic symbol information. CONSTITUTION:A book data base 5 stores phonetic symbol information including vocal sound information and rhythm information. The center system 1 transmits phonetic symbol information stored in the book data base 5 to the terminal 2, and the terminal 2 receives the transmitted phonetic symbol information. Thus, phonetic symbol information stored in the book data base 5 can be supplied to a user with a synthetic sound by synthesizing sound based on the phonetic symbol information. Thus, the operability of the system is improved and the physically handicapped user can easily use the system.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention relates to information about publications such as newspapers, magazines, books, music information, travel information,
The present invention relates to an information access system suitable for use in accessing a center system having a database storing house information and the like from a terminal.

[0002]

2. Description of the Related Art Conventionally, as an information access system that allows a user to access a database of a center system from home or the like, there is, for example, a house information facsimile service system including house rental information and sales information. In this system, the details of the house information described in the house information magazine are stored in the database of the center system, and the user accesses the center system via, for example, a telephone line, and the information number of the desired house information. By operating the push button or the like, the details of the desired housing information can be obtained from the facsimile of the desired telephone number.

As an information access system, for example, a depositor such as a bank uses a telephone (via a telephone line).
There is also a system in which the balance inquiry can be performed by accessing the center system of the bank and inputting the own account number by operating a push button or the like.

Further, there is also an information access system in which a desired book can be searched by inputting the date of publication, the author, the publisher, the keyword, etc., for example.

[0005]

In the information access system as described above, desired information is operated by operating a terminal, that is, operating push buttons of a telephone or a facsimile or operating a keyboard. You are able to get.

However, the operation of the push buttons and the keyboard has been difficult for users with physical disabilities, especially for visually impaired users and users who cannot freely move their limbs.

The present invention has been made in view of such a situation, and is to improve the operability of the system and to provide a system that is easy to use even for a user with a physical disability. is there.

[0008]

An information access system according to claim 1 is connected with a center system having a database, such as a book database 5, which stores information, and the center system 1 by a communication line.
Terminal 2 to access information in the book database 5
In the information access system consisting of, the book database 5 stores at least phonetic symbol information including phonological information and prosody information, and the center system 1 transfers the phonetic symbol information stored in the book database 5 to the terminal 2. It is characterized in that the terminal 2 receives the phonetic symbol information transmitted from the center system 1.

According to another aspect of the information access system of the present invention, the terminal 2 comprises a voice synthesizing device 2-6 as a voice synthesizing means for synthesizing a voice from phonetic symbol information.

An information access system according to a third aspect is characterized in that the phonetic symbol information relates to a publication.

In the information access system according to the fourth aspect, the terminal 2 is a microphone 2-21 as an input means for inputting a keyword of a desired publication, and a keyboard 2
-3, button section 2-4, or jog shuttle section 2-5
Further, the keyword output from the microphone 2-21, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 is transmitted to the center system 1, and the center system 1 transmits from the terminal 2. Search unit 3 as a search means for searching phonetic symbol information on a publication corresponding to the selected keyword from the book database 5.
-6 is further provided, and the search result of the search unit 3-6 is transmitted to the terminal 2.

In the information access system according to claim 5, the center system 1 transmits, to the terminal 2, the phonetic symbol information relating to the publication, which is the search result of the search unit 3-6, regarding the outline thereof. Is characterized by.

In the information access system according to the sixth aspect, the keyword of the desired publication is input to the microphone 2-21 by voice, and the center system 1 recognizes the voice input to the microphone 2-21. Further, a recognition unit 3-4 as a voice recognition means for
It is characterized in that the phonetic symbol information regarding the publication is searched from the book database 5 based on the recognition result of the recognition unit 3-4.

In the information access system according to claim 7, when the recognition unit 3-4 inputs a plurality of keywords into the microphone 2-21 by voice, a combination of voice recognition results of the plurality of keywords is combined. It is characterized in that it determines whether or not there is a possibility, and outputs the speech recognition result of each of the plurality of keywords based on the determination result.

The information access system according to claim 8 is characterized in that the input means is a keyboard, a button, a jog dial, or a shuttle ring.

In the information access system according to claim 9, the terminal 2 is an information recording / reproducing apparatus as recording means for recording phonetic symbol information on a recording medium such as a magnetic disk, a magnetic tape, an optical disk or a magneto-optical disk. 2-12 is further provided.

In the information access system according to the tenth aspect, the terminal 2 is an information recording / reproducing apparatus 2-1 as a reproducing means for reproducing the phonetic symbol information recorded on the recording medium.
2 is further provided.

In the information access system according to the eleventh aspect of the present invention, the processing step S2- of the program as the mark addition means for the terminal 2 to add the mark to the predetermined position of the synthesized sound output from the voice synthesizer 2-6. The speech synthesizer 2-6 is further provided with 10, and is characterized in that the synthesized speech is resynthesized and outputted from the position where the mark is added.

The information access system according to claim 12 is an information access system comprising a center system 1 and a terminal 2 connected to the center system 1 by a communication line, wherein the terminal 2 inputs a voice input means. 2-21 and microphone 2 as
And an analysis unit 2-23 as an acoustic analysis unit that acoustically analyzes the voice input to -21. The acoustic analysis result of the analysis unit 2-23 is transmitted to the center system 1, and the center system 1 transmits the result to the terminal 2. Analysis section 2-2 sent from
Based on the acoustic analysis result of No. 3, the recognition unit 3-4 as the center side voice recognition means for performing voice recognition, and the recognition unit 3-4.
And a system control unit 3-3 as a center side control unit that performs a predetermined process in accordance with the voice recognition result.

The information access system according to a thirteenth aspect is characterized in that the system control section 3-3 transmits the voice recognition result of the recognition section 3-4 to the terminal 2.

In the information access system according to claim 14, the recognition unit 3-4 performs voice recognition by dictation, and the terminal 2 recognizes the voice of the recognition unit 3-4 transmitted from the system control unit 3-3. Of the results, a voice recognition device 2-2 as a voice recognition result correction means for correcting the erroneously recognized part, a keyboard 2-3, a button section 2-
4 or a jog shuttle unit 2-5 is further provided, and a voice recognition device 2-2, a keyboard 2-3, a button unit 2-4,
Alternatively, the correction result by the jog shuttle unit 2-5 is transmitted to the center system 1.

An information access system according to a fifteenth aspect is characterized in that the voice recognition result correction means is a jog dial or a shuttle ring.

In the information access system according to the sixteenth aspect, the recognition unit 3-4 performs voice recognition by dictation, and the voice recognition result of the recognition unit 3-4 transmitted from the system control unit 3-3 is incorrect. If so, microphone 2-2
1, the voice is re-input, and the recognition unit 3-4 causes the microphone 2 to
It is characterized in that the voice re-inputted to -21 is again voice-recognized based on the incorrect voice recognition result.

In the information access system according to the seventeenth aspect, the terminal 2 is a voice as a voice synthesizing means for synthesizing a voice corresponding to the voice recognition result of the recognition unit 3-4 transmitted from the system control unit 3-3. It is characterized by further comprising a synthesizer 2-6.

In the information access system according to claim 18, the terminal 2 recognizes, based on the acoustic analysis result of the analysis section 2-23, a recognition section 2-24 as a terminal side voice recognition means, and recognition. It is characterized by further comprising a central control device 2-1 as a terminal side control means for performing a predetermined process in response to the voice recognition result of the section 2-24.

In the information access system according to the nineteenth aspect, the center system 1 further includes a book database 5 as a database storing information about publications, and the system control unit 3-3 and the recognition unit 3-. Information about publications is searched from the book database 5 and transmitted to the terminal 2 in accordance with the voice recognition result of No. 4.

[0027] In the information access system according to the twentieth aspect, the terminal 2 synthesizes the voice corresponding to the information on the publication transmitted by the system control unit 3-3, as a voice synthesizing device 2-6. Is further provided.

In the information access system according to the twenty-first aspect, the terminal 2 transmits the range of the vocabulary to be subjected to the voice recognition to the center system 1, and the recognition unit 3-4 transmits the vocabulary transmitted from the terminal 2. The feature is that voice recognition is performed in the range of.

The information access system according to a twenty-second aspect is characterized in that the terminal 2 compresses the acoustic analysis result of the analysis section 2-23 and transmits it to the center system 1.

In the information access system according to the twenty-third aspect, the terminal 2 outputs the information about the publication transmitted by the system control unit 3-3 to the display 2-7 as a braille display or the printer 2 as a braille printer. -8 is further provided.

An information access system according to a twenty-fourth aspect is characterized in that the voice is uttered in word units or syllable units.

In the information access system according to the twenty-fifth aspect, the terminal 2 outputs the synthesized sound synthesized by the voice synthesizer 2-6 to a recording medium such as a magnetic tape, a magnetic disk, an optical disk or a magneto-optical disk. The audio recording / reproducing apparatus 2-13 as recording means for recording is further provided.

In the information access system according to the twenty-sixth aspect of the present invention, the terminal 2 uses the voice recognition device 2-2 as the synthetic voice control means for controlling the synthetic voice which the voice synthesizer 2-6 synthesizes and outputs, and the keyboard 2. -3, a button portion 2-4, or a jog shuttle portion 2-5 is further provided.

In the information access system according to the twenty-seventh aspect, the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 controls the voice quality of the synthesized voice. Characterize. The information access system according to claim 28 is the voice recognition device 2-2, the keyboard 2-3, the button unit 2-4, or the jog shuttle unit 2.
-5 controls the speech rate of the synthetic sound.

In the information access system according to claim 29, the voice recognition device 2-2, the keyboard 2-3, the button part 2-4, or the jog shuttle part 2-5 controls the pitch of the synthesized voice. It is characterized by doing.

An information access system according to a thirtieth aspect is characterized in that the synthesized voice control means is a jog dial or a shuttle ring.

An information access system according to a thirty-first aspect is characterized in that the terminal 2 further comprises an audio recording / reproducing device 2-13 as a reproducing means for reproducing the synthetic sound recorded on the recording medium.

An information access system according to a thirty-second aspect is characterized in that the voice recording / reproducing apparatus 2-13 reversely reproduces the synthetic sound.

In the information access system according to a thirty-third aspect, the voice synthesizing device 2-6 corrects the synthesized sound in accordance with a predetermined auditory characteristic, as a characteristic correcting means, for the hearing-impaired person. It is characterized by comprising a frequency correction unit 2-66 for the deaf.

An information access system according to a thirty-fourth aspect is characterized in that a predetermined auditory characteristic can be changed.

The information access system according to the thirty-fifth aspect is characterized in that the correction unit 2-64 for the hearing impaired or the frequency correction unit 2-66 for the deaf is a filter.

A recording medium according to a thirty-sixth aspect is characterized in that phonetic symbol information including phoneme information and prosody information is recorded.

An information access system according to a thirty-seventh aspect is characterized in that the center system 1 further comprises an authoring system 4 which creates phonetic symbol information from information on publications and stores it in the book database 5.

The information access system according to a thirty-eighth aspect is characterized in that the authoring system 4 comprises a phonetic symbol string editing unit 4-5 as an editing means for editing phonetic symbol information.

The information access system according to a thirty-ninth aspect is characterized in that the authoring system 4 includes a rule synthesizing section 4-77 as a voice synthesizing means for synthesizing a voice from phonetic symbol information.

In the information access system according to claim 40, the authoring system 4 analyzes the text data of the publication and, based on the analysis result, generates the phonetic symbol information. 3 is provided.

In the information access system according to the forty-first aspect, the authoring system 4 includes the text analysis unit 4-3.
If the analysis result of 1) includes a reading error of a word in the text data of the publication, the word reading correction unit 4 as a reading error correction unit that corrects the reading error of the word.
-71 is further provided.

In the information access system according to claim 42, the authoring system 4 includes the text analysis unit 4-3.
If the vowel devoicing error in the text data of the publication is found in the analysis result of, the phonological unvoicing / voicing correction unit as a voicing error correction means for correcting the vowel devoicing error. 4-73 is further provided.

In the information access system according to the forty-third aspect, the authoring system 4 includes the text analysis section 4-3.
When there is an error in the accent of a word in the text data of the publication in the analysis result of 1., an accent correction unit 4-75 is further provided as an accent error correction unit for correcting the error in the accent of the word. And

In the information access system according to the forty-fourth aspect, the authoring system 4 includes the text analysis unit 4-3.
If there is an error in the position of the pose inserted in the text data of the publication in the analysis result of, the pose correction unit 4-74 as a pose position error correction means for correcting the error in the position of the pose is further added. It is characterized by being provided.

In the information access system according to the forty-fifth aspect, the authoring system 4 includes the text analysis unit 4-3.
If there is an error in the phrase in the text data of the publication in the analysis result of 1., the phrase correction unit 4-as a phrase error correction means for correcting the error in the phrase.
76 is further provided.

In the information access system according to the forty-sixth aspect, the authoring system 4 includes the text analysis unit 4-3.
If there is an error in vowel lengthening in the text data of the publication, the vowel lengthening correction unit 4-72 as a lengthening error correction unit that corrects the vowel lengthening error Is further provided.

In the information access system according to claim 47, the authoring system 4 has a voice analysis unit 4-78 as a voice input means for inputting voice and a voice analysis unit 4-.
The prosody information analysis unit 4-79 and the prosody information analysis unit 4- as extraction means for extracting the prosody information of the voice input to 78.
It further comprises an editing function unit 4-70 as a correction means for correcting an error in the analysis result of the text analysis unit 4-3 based on the prosody information of the voice extracted by 79.

In the information access system according to the forty-eighth aspect, the authoring system 4 further comprises an OCR (optical character reading device) 4-12 as a character reading means for reading a character from a publication and generating text data. It is characterized by

In the information access system according to a forty-ninth aspect, the authoring system 4 synthesizes a voice from the output of the OCR 4-12 by the rule synthesizing section 4 as a voice synthesizing means.
-77 and an error in reading characters from a publication by the OCR 4-12 are corrected based on the synthesized sound output from the rule synthesis unit 4-77.
-70 is further provided.

In the information access system according to claim 50, the authoring system 4 converts the graphic data of the publication into text data which describes the graphic data, and the text analysis unit 4-3, the text which describes the graphic data. Characterized by further analysis of the data.

The information access system according to a fifty-first aspect is characterized in that the text data of the publication includes information about an outline of the description items of the publication.

In the information access system according to claim 52, the authoring system 4 uses the keyboard as a summary input means for inputting the summary of matters described in the publication as text data of the publication to the text analysis unit 4-3. 4-
7 is further provided.

In the information access system according to claim 53, the center system 1 prepares the quick-reading information for quick-reading the publication, and the quick-reading / text search information creating section 4- as the quick-reading information creating means. 4, the book database 5 stores the fast-reading information together with the phonetic symbol information about the publication, and the speech synthesizer 2-6 produces a synthetic sound for quick-reading the publication based on the fast-reading information. It is characterized by outputting.

The information access system according to claim 54 is characterized in that the fast-reading information is a keyword extracted from a publication.

The information access system according to the forty-fifth aspect is characterized in that the quick-reading information is the table of contents information of the publication.

The information access system according to claim 56 is characterized in that the center system 1 further comprises a keyboard 4-7 as a table of contents creating means for creating table of contents information of the publication.

In the information access system according to claim 57, the keyword is attribute data representing the attributes of the publication, and the book database 5 has a correspondence relationship between the attribute data and a correspondence relationship between the attribute data and the publication. Is stored, and the search unit 3-6 searches with reference to the relationship table.

In the information access system according to claim 58, the publication or attribute data is provided with a unique publication identification number or attribute identification number for identifying the publication or attribute data. It is characterized in that the correspondence is described by the number and the attribute identification number.

In the information access system according to claim 59, the center system 1 further includes a search memory 3-42 as a past search information storage means for storing a publication identification number or an attribute identification number retrieved in the past. The search unit 3-6 is characterized by first performing a search by referring to the search memory 3-42.

In the information access system according to claim 60, the center system 1 further comprises a personal database 3-44 as a search range storing means for storing the search range, and the search section 3-6 comprises a personal database 3-44. It is characterized in that a search is performed within the search range stored in 44.

In the information access system according to claim 61, the book utilization table 3 as a retrieval frequency storing means for the center system 1 to store the frequency with which the phonetic symbol information regarding the publication is retrieved by the retrieval unit 3-6. With -43,
It is characterized in that the retrieval unit 3-6 refers to the book usage degree table 3-43 and sequentially retrieves from the group of phonetic symbol information relating to frequently published publications.

In the information access system according to the forty-sixth aspect, the attribute data is voice-recognized, and the search unit 3-6 produces the pronunciation of a publication corresponding to the attribute data having a high likelihood of the voice recognition result. The feature is that the search is sequentially performed from the symbol information.

[0069] In the information access system according to claim 63, the input means is a display 2-7 as a keyword output means for outputting up to a predetermined number of characters among the keywords, and a keyword from the display 2-7. A shuttle ring 2-51 as an output control means for controlling the output, and a confirm button 2-38 as a selection means for selecting a desired one of the keywords output from the display 2-7 are provided. To do.

An information access system according to a sixty-fourth aspect is characterized in that the attribute data is a title of a publication, a publisher name, an author name, a publication date, the number of pages, or a field.

In the information access system according to a sixty-fifth aspect, the terminal 2 has a broadcast receiving device 2-10 as a broadcast receiving means for receiving an emergency broadcast regarding emergency alert, and a display 2 as a notifying means for notifying the emergency alert. 7
And is further provided.

The information access system according to the sixty-sixth aspect is characterized in that the center system 1 and the terminal 2 perform wireless communication.

An information access system according to a sixty-seventh aspect is characterized in that the terminal 2 operates using a battery as a power source.

In the information access system according to claim 68, the center system 1 is connected to another center system via a network, and serves as a network control means for controlling the exchange of information with the other center system. The network control unit 6 is further provided.

The information access system according to claim 69 is characterized in that the center system 1 is connected to a plurality of terminals.

[0076]

In the information access system of the present invention, the book database 5 stores at least phonetic symbol information including phonological information and prosody information, and the center system 1
The phonetic symbol information stored in the book database 5,
The phonetic symbol information transmitted from the center system 1 is transmitted to the terminal 2, and the terminal 2 receives the phonetic symbol information transmitted from the center system 1. Therefore, by performing the voice synthesis based on the phonetic symbol information, the information stored in the book database 5 can be provided to the user by the synthesized voice. Further, in this case, since the phonetic symbol information is transmitted / received, the amount of data to be transmitted / received is smaller than that in the case where voice itself is transmitted / received, and the communication cost can be reduced.

Further, in the information access system of the present invention, the terminal 2 is the microphone 2-21, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2.
The keyword entered in -5 is transmitted to the center system 1. Then, the center system 1 searches the phonetic symbol information regarding the publication corresponding to the keyword transmitted from the terminal 2 from the book database 5, and transmits the retrieval result to the terminal 2. Therefore, by performing voice synthesis based on the phonetic symbol information as the search result, it becomes possible to provide the user with the searched information by the synthesized voice.

Further, in the information access system of the present invention, the keyword of the desired publication is input by voice to the microphone 2-21. Then, in the center system 1, the voice input to the microphone 2-21 is recognized, and the phonetic symbol information regarding the publication is searched from the book database 5 based on the recognition result. Therefore, desired information can be obtained only by inputting a keyword by voice without operating a keyboard or the like.

Further, in the information access system of the present invention, when a plurality of keywords are input by voice to the microphone 2-21, the recognition unit 3-4 causes the recognition unit 3-4 to combine the voice recognition results of the plurality of keywords. It is determined whether or not there is a possibility, and the voice recognition result of each of the plurality of keywords is output based on the determination result. Therefore, the false recognition rate can be reduced.

Further, in the information access system of the present invention, a mark is added to a predetermined position of the synthesized sound output from the voice synthesizer 2-6, and the position where the mark is added by the voice synthesizer 2-6. From, the synthesized sound is re-synthesized and output. Therefore, the synthesized sound of the desired portion can be immediately obtained.

In the information access system of the present invention, the voice input to the microphone 2-21 is acoustically analyzed at the terminal 2, and the acoustic analysis result is transmitted to the center system 1. Then, in the center system 1, voice recognition is performed based on the acoustic analysis result transmitted from the terminal 2, and predetermined processing is performed corresponding to the voice recognition result. Therefore, as compared with the case where the audio signal itself is transmitted to the center system 1, the deterioration of the audio can be prevented and the communication amount can be reduced. Further, in this case, since a large-scale voice recognition device is not required on the terminal 2 side, the terminal 2 can be made compact.

Further, in the information access system of the present invention, the system control section 3-3 transmits the voice recognition result of the recognition section 3-4 to the terminal 2. Therefore, the operation control of the terminal 2 by voice can be performed without providing the voice recognition device on the terminal 2 side.

Furthermore, in the information access system of the present invention, when the voice recognition result is incorrect, the microphone 2-
The voice is re-input to 21 and the voice is recognized again based on the incorrect voice recognition result. Therefore, the same voice is prevented from being erroneously recognized again.

Further, in the information access system of the present invention, the voice corresponding to the voice recognition result of the recognition unit 3-4 transmitted from the system control unit 3-3 to the terminal 2 is synthesized. Therefore, even a visually impaired user can check the voice recognition result.

Further, in the information access system of the present invention, the system control unit 3-3 searches the book database 5 for the information on the publication corresponding to the voice recognition result of the recognition unit 3-4, and then the terminal Send to 2. Therefore, desired information can be obtained by voice without operating a keyboard or the like.

In the information access system of the present invention, the terminal 2 synthesizes the voice corresponding to the information about the publication transmitted by the system control unit 3-3. Therefore, even the visually impaired user can confirm the contents of the publication.

Further, in the information access system of the present invention, the voice synthesizer 2-6 corrects the synthesized voice in accordance with a predetermined auditory characteristic. Therefore, it is possible to provide a synthesized sound that is easy to hear to a deaf user.

In the recording medium of the present invention, phonetic symbol information including phonological information and prosody information is recorded. Therefore, the synthesized sound can be easily obtained.

In the information access system of the present invention, the authoring system 4 creates phonetic symbol information from information on publications and stores it in the book database 5. Therefore, by generating the synthetic sound from the phonetic symbol information, it becomes possible to provide the information of the publication by voice.

Further, in the information access system of the present invention, the rule synthesizing unit 4-77 synthesizes the voice from the phonetic symbol information, so that the phonetic symbol information created by the authoring system 4 can be confirmed.

Further, in the information access system of the present invention, the prosody information of the voice input to the voice analysis unit 4-78 is extracted, and the error of the analysis result of the text analysis unit 4-3 is based on this prosody information. To correct. Therefore, the error in the analysis result can be corrected by voice without operating the keyboard or the like.

Further, in the information access system of the present invention, the quick reading / text search information creating unit 4-4 creates quick reading information for quick reading of the publication.
The book database 5 stores fast-reading information together with phonetic symbol information about publications. Then, the voice synthesizer 2-6 outputs a synthetic sound for early reading of the publication based on the fast reading information. Therefore, the content of the publication
You will be able to grasp quickly.

In the information access system of the present invention, the keywords are attribute data representing attributes of publications, and the book database 5 represents correspondence between attribute data and correspondence between attribute data and publications. The relationship table is stored, and the search unit 3-6 searches by referring to the relationship table. Therefore, the search can be performed quickly.

In the information access system of the present invention, the publication or attribute data is provided with a unique publication identification number or attribute identification number for identifying the publication or attribute data, and the relational table shows the publication identification number. The correspondence is described by the number and the attribute identification number. Therefore, the search can be performed more quickly.

Further, in the information access system of the present invention, the search memory 3-42 stores the publication identification number or the attribute identification number retrieved in the past, and the retrieval unit 3-6 first retrieves it. A search is performed by referring to the memory 3-42. Therefore, when a search that was performed in the past is searched again, the search result can be immediately obtained.

Further, in the information access system of the present invention, the search range is stored in the personal database 3-44, and the search unit 3-6 searches within the search range stored in the personal database 3-44. . Therefore, the search can be performed quickly.

Further, in the information access system of the present invention, the frequency of retrieval of phonetic symbol information relating to the publication by the retrieval unit 3-6 is stored in the book utilization degree table 3-43, and the retrieval unit 3-6 stores the frequency. , The book utilization table 3-43 is referred to, and the phonetic symbol information related to the frequently used publications is searched in sequence. Therefore, it is possible to immediately search for information about publications that have been searched frequently in the past.

In the information access system of the present invention, the keywords up to a predetermined number of characters among the keywords are output to the display 2-7 while being controlled by the shuttle ring 2-51. And display 2
Of the keywords output to -7, the desired one is
It is selected by the confirm button 2-38. Therefore, the desired keyword can be easily selected.

In the information access system of the present invention, the center system 1 is connected to another center system via a network, and the network control unit 6 controls the exchange of information with the other center system. . Therefore, it becomes possible to share information among a plurality of center systems.

[0100]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the information access system of the present invention. This information access system is composed of a center system 1 and a terminal 2 which is connected to the center system 1 by a telephone line such as PSTN or ISDN and which can communicate with the center system 1.

The center system 1 and the terminal 2
And can communicate wirelessly. Further, in FIG. 1, only one terminal 2 is connected to the center system 1, but a plurality of terminals can be connected to the center system 1. Furthermore, the terminal 2 is designed not only to be powered by an outlet but also to be operated by using a battery as a power source, so that it can be used not only at home but also on the go. Has been done.

The terminal 2 uses the voice input type database retrieval / transmission system 3 of the center system 1 as an interface to a book database 5 in which information on publications such as library information, newspaper information, magazine information and new publication information is stored. You can access it and get the information you want. Note that the book database 5 stores phonetic symbol information including phonological information and prosodic information in addition to the text data as publication information, and the terminal 2 receives this information and publishes it. In addition to displaying and printing the text data of an object, a synthesized sound is generated from phonetic symbol information, and a voice reading a publication is thereby obtained.

The center system 1 includes a voice input database search / transmission system 3, an authoring system 4, a book database 5, and a network control unit 6.
It consists of The authoring system 4 is configured to store the information of the publication in the book database 5, create a database of the information of the publication, and edit the information stored in the book database 5. The network control unit 6 controls the communication between the center system 1 and other center systems, thereby making it possible to share the library database 5 of the center system 1 with the databases of other center systems. ing.

Details of each block will be described below.

First, FIG. 2 shows a detailed configuration of the network control unit 6. The network control unit 6 controls a database control unit 6-1 that controls data exchange with the book database 5, a CPU 6-2 that controls operation of the network control unit 6, and communication that controls data exchange with another center system. The control unit 6-3 and a book ID database 6-4 storing a book ID (details will be described later) as a unique number assigned to a publication stored in the book database 5 are included.

Now, for example, as shown in FIG. 3, the center system 1 is replaced with other center systems 1-1 to 1-7.
Consider a case where a network is configured by connecting in a ring shape with. In this case, the CPU 6-2 performs a predetermined process, for example, the process shown in the flowchart of FIG. 4 in order to share the book database 5 of the center system 1 with the databases of the other center systems 1-1 to 1-7. It is performed every period (for example, every one week).

That is, first, in step S1-1, the book ID attached to the publication stored in the book database 5 is read out via the database control unit 6-1 and stored in the book ID database 6-4. Book ID
Compared to. Then, the process proceeds to step S1-2, the book ID stored in the book ID database 6-4 and the book I attached to the publication stored in the book database 5.
It is determined whether or not D matches. Step S1-2
If it is determined that the book ID stored in the book ID database 6-4 and the book ID attached to the publication stored in the book database 5 match each other, skip steps S1-3 to S1-5. And step S
Go to 1-6.

In step S1-2, the book I
The book ID stored in the D database 6-4 and the book ID attached to the publication stored in the book database 5
If it is determined that does not match, that is, if the information of the publication is newly made into a database by the authoring system 4 (FIG. 1) and this information is additionally stored in the book database 5, the step Proceeding to S1-3, the book ID of the publication (hereinafter referred to as additional publication) additionally stored in the book database 5 is additionally stored in the book ID database 6-4.

Then, the process proceeds to step S1-4, and the subsequent center system (here, the center system 1 adjacent to the center system 1 in the clockwise direction, for example, is used.
-1 (meaning FIG. 3) is checked (determined) via the communication control unit 6-3 as to whether or not data reception is possible. When it is determined in step S1-4 that the subsequent center system (center system 1-1) is not in a data receivable state, the process returns to step S1-4. If it is determined in step S1-4 that the subsequent center system is in a state where data can be received, the process proceeds to step S1-5, and the information of the additional publication is stored in the communication control unit 6-3. To the subsequent center system, and the process proceeds to step S1-6.

In step S1-6, the previous center system (here, the center system 1-7 which is adjacent to the center system 1 in the counterclockwise direction, for example) is used.
(Means FIG. 3) is enabled to receive data, the process proceeds to step S1-7, and it is determined whether data has been transmitted from the previous center system (center system 1-7). It

If it is determined in step S1-7 that the previous center system has not transmitted data, the process returns to step S1-7. Also, step S1-
When it is determined that the data is transmitted from the previous center system in step 7, the process proceeds to step S1-8, the data transmitted from the previous center system is received via the communication control unit 6-3, and The book ID of the data,
The book ID stored in the book database 6-4 is compared.

Then, as a result of the comparison, among the book IDs of the data from the previous center system, those which are not stored in the book database 6-4 are additionally written in the book ID database 6-4, and this addition is made. The data corresponding to the written book ID is supplied to the book database 5 via the database control unit 6-1 and stored therein, and the process ends.

On the other hand, in the center systems 1-1 to 1-7 (FIG. 3) other than the center system 1, step S1
The process is started from -6, and after the process of step S1-8 is performed, the processes of steps S1-1 to S1-5 are sequentially performed, and the process is ended.

As described above, the data (information) stored in the book database 5 is updated, for example, on a weekly basis so that the database can be shared.

The data sharing process can be performed not only on the added publication information but also on the deleted publication information.

Further, it is preferable that the updating process is performed at a time when there is no user who accesses the center system 1, for example, at night on a holiday.

Next, FIG. 5 shows the terminal 2 shown in FIG.
The detailed configuration of is shown. The central control device 2-1 controls each block constituting the terminal 2, and also corresponds to the output signals of the voice recognition device 2-2, the keyboard 2-3, the button part 2-4, and the jog shuttle part 2-5. Then, predetermined processing is performed. For example, as shown in FIG. 6, the voice recognition device 2-2 includes a microphone 2-21, an A / D converter 2-22, an analysis unit 2-23, a recognition unit 2-24, and a memory 2-25.
According to the control of the central control unit 2-1, a voice input as a command or other necessary information from the microphone 2-21 is acoustically analyzed, and voice recognition is further performed if necessary. .

That is, the microphone 2-21 converts the input voice into a voice signal as an electric signal and outputs it to the A / D converter 2-22. The A / D converter 2-22 performs A / D conversion on the audio signal from the microphone 2-21 at a predetermined sampling frequency and outputs it to the analysis unit 2-23. The analysis unit 2-23 acoustically analyzes the audio signal from the A / D converter 2-22 and extracts a characteristic parameter (acoustic parameter) of the audio such as a linear prediction coefficient and a cepstrum coefficient. Further, in the analysis unit 2-23, the central control unit 2-
According to the control of No. 1, the acoustic parameter extracted from the audio signal is transmitted to the center system 1, or is subjected to, for example, vector quantization, and is supplied to the recognition unit 2-24.

The acoustic parameters transmitted to the center system 1 are recognized in the voice input type database search / transmission system of the center system 1 as described later.

Further, when transmitting the acoustic parameters to the center system 1, the central controller 2-1 can compress the acoustic parameters and transmit the compressed acoustic parameters. In this case, in the center system 1, the compressed acoustic parameter is expanded and then processed.

Further, when transmitting the acoustic parameters to the center system 1 for voice recognition, the central control unit 2-1 transmits the limitation information limiting the range of the vocabulary to be recognized together with the acoustic parameters. Is made possible. In this case, the center system 1 performs voice recognition within the range of vocabulary based on the limited information.

The memory 2-25 stores, for example, a phoneme-based model and a dictionary of words as a sequence of the phonemes (word dictionary) obtained by merging learning, and the recognition unit 2-24. Then, the probability that the acoustic parameter sequence output from the analysis unit 2-23 is observed from the model stored in the memory 2-25 is, for example, HMM (Hidden M
The speech input to the microphone 2-21 is recognized on a phoneme-by-phoneme basis by being calculated based on the arkov Models) method or the like.

Here, the microphone 2-21 is normally input with a voice as a response to an inquiry (question) from the central control unit 2-1. Therefore, the central controller 2
At -1, the voice input to the microphone 2-21 can be expected.

Therefore, during voice recognition, a recognition control signal for controlling the range of words to be recognized is output from the central control device 2-1 to the recognition unit 2-24. 2-24 refers to a word in the range indicated by the recognition control signal among the words described in the word dictionary stored in the memory 2-25, obtains a sequence of recognition results in phoneme units, that is, obtains a word, The likelihood is output to the central controller 2-1.

The voice recognition can be performed according to a voice recognition algorithm other than the HMM method.

In the central control device 2-1, the word (or word string) as the voice recognition result supplied from the voice recognition device 2-2 as described above is converted into the above-mentioned expected response and word likelihood. Based on the degree, the sound left after the selection is finally understood as the sound input to the microphone 2-21.

Returning to FIG. 5, the keyboard 2-3, the button section 2-4, and the jog shuttle section 2-5 are operated when a command and other necessary information are input. The button portion 2-4 and the jog shuttle portion 2-5 are integrally formed on the panel, as shown in FIG. 7, for example.

That is, the button portion 2-4 is Yes / confirmation button 2-31, No button 2-32, left button 2-3.
3, right button 2-34, recognition method switching button 2-36,
Read button 2-37, confirm button 2-38, display button 2-39, setting button 2-40, and output button 2-
41, etc., the jog shuttle unit 2-5
Is shuttle ring 2-51 and shuttle ring 2
It is composed of a play button 2-52 and a stop button 2-53 arranged in -51. The jog shuttle unit 2-5 can be configured by replacing the shuttle ring 2-51 with a jog dial. Further, the jog shuttle unit 2-5 may be configured by providing both the shuttle ring 2-51 and the jog dial.

Here, in the terminal 2, input of commands and other necessary information to the central control unit 2-1 is performed by operating the above-mentioned keyboard 2-3, button section 2-4, and jog shuttle section 2-5. not only,
Microphone 2 constituting the above-mentioned voice recognition device 2-2 (FIG. 6)
It is also possible to do this by inputting a voice corresponding to a command or other necessary information from -21.

Next, the speech synthesizer 2-6 constituting the terminal 2 (FIG. 5) is constructed, for example, as shown in FIG. In the reading information analysis unit 61, information about how to read the publication as well as phonetic symbol information from the center system 1 (for example, information about voice quality and speed during reading).
(Hereinafter referred to as reading information) is supplied via the central control device 2-1. The reading information analysis unit 61 separates the phonetic symbol information into phonological information and prosody information, further analyzes the reading information, and outputs the reading information together with the phonological information and the prosody information to the phonological / prosodic information generating unit 2-62.

In the phonological / prosodic information generation unit 2-62, the speech unit connection unit 2-65 is controlled based on the analysis result of the phonological information, the prosody information, and the recitation information. When text data, not phonetic symbol information, is input to the speech synthesizer 2-6, the reading information analysis unit 2-61.
From that, the text data is output as it is to the phonological / prosodic information generation unit 2-62. In this case, the phonological / prosodic information generation unit 2-62 outputs the phonological information from the text data, It is designed to generate prosody information.

On the other hand, the voice unit data storage unit 2-63 stores voice unit data (eg, digitized waveform data) necessary for regular voice synthesis in units such as CV and CV / VC. This speech unit data is read by the speech unit connecting unit 2-65 via the hearing-impaired person correcting unit 2-64.

Here, in the correction unit 2-64 for the hearing impaired, D
The formant of the voice unit data is emphasized as necessary so that the synthesized sound output from the / A converter 2-67 becomes easy to hear.

In the speech unit connection unit 2-65, the speech unit data storage unit 2-6 is controlled by the phoneme / prosodic information generation unit 2-62 via the deafening person correction unit 2-64.
Necessary speech unit data is read from No. 3, and the read speech unit data is connected so as to form a continuous speech waveform. This voice waveform (synthetic voice signal) is output to the frequency correction unit 2-66 for the hearing impaired. The frequency correction unit 2-66 for the hearing-impaired person is composed of, for example, a low-frequency or high-frequency emphasis digital filter having a predetermined number of taps, and makes the synthesized voice signal from the phoneme unit connection unit 2-65 easy to hear. So that the spectrum is emphasized and output to the D / A conversion unit 2-67. In the D / A conversion unit 2-67, the synthesized voice signal from the frequency correction unit 2-66 for the hearing impaired is
A-converted into an analog signal, which is output as a synthesized sound from a speaker (not shown), for example, a speaker.

The voice quality and utterance speed (speech speed) of the synthesized sound as the recitation information, the formant emphasis characteristic in the correction unit 2-64 for the deaf and the frequency emphasis characteristic in the frequency correction unit 2-66 for the deaf are as follows. For example, the keyboard 2-3 or the right button 2-3 of the button section 2-4 (FIG. 7).
3 and left button 2-34, jog shuttle unit 2-5
It can be changed by operating the shuttle ring (FIG. 7).

Also, these can be changed by inputting a voice from the voice recognition device 2-2 (for example, by uttering "higher utterance speed" or the like, the utterance speed of the synthetic sound can be increased. It can be done).

Returning to FIG. 5, the display 2-7 displays the information transmitted from the center system 1 under the control of the central control unit 2-1 and the information (question (question (question) )) Is displayed. The display 2-7 can be a braille display. The printer 2-8 is a central control unit 2-
According to the control of No. 1, the information transmitted from the center system 1 is printed out. The printer 2-8 can be a Braille printer.

The communication device 2-9 is, for example, a modem as an interface between the center system 1 and the central control device 2-1 and controls communication between them. The broadcast receiving device 2-10 is configured similarly to, for example, a radio or a television receiver, and normally receives a radio program (radio broadcast) or a television program (television broadcast), and receives the received program, Display 2
-7 or output from a speaker.
Further, the broadcast receiving device 2-10 is adapted to be able to always receive the emergency broadcast, and even if the power switch (not shown) of the terminal 2 is turned off when the emergency broadcast is received. The contents of the emergency broadcast are displayed on the display 2-7 or output from the speaker.

The timer circuit 2-11 is adapted to measure the time, and causes the central control unit 2-1 to refer to the measured time, for example, at a time when the communication charge is low (for example, at midnight). Let's access the center system 1,
The information retrieved from the book database 5 (FIG. 1) can be received when accessed during the day.

The information recording / reproducing apparatus 2-12 is, for example, a semiconductor memory device, magnetic disk device (floppy disk device, hard disk device), magneto-optical disk device (mini disk device), optical disk device (CD-ROM).
Device, etc., for example, information transmitted from the center system 1 and other information necessary for the operation of the device are recorded media (for example, semiconductor memory, magnetic disk,
Recording on a magneto-optical disk, optical disk, etc. Also,
The information recording / reproducing device 2-12 reproduces the information recorded on the recording medium according to the control of the central controller 2-1.

The voice recording / reproducing device 2-13 is, for example, an audio cassette tape recorder, a digital audio tape (DAT) device, a mini disk device, or the like, and the voice (synthesized sound) synthesized by the voice synthesizer 2-6.
Is recorded on a recording medium (for example, a cassette tape, a digital audio tape, a mini disk, etc.). Further, the audio recording / reproducing device 2-13 reproduces or reversely reproduces the synthetic sound recorded on the recording medium according to the control of the central controller 2-1. It should be noted that playback of the synthetic sound, stop of playback,
Alternatively, for fast forward (including double speed reproduction) and rewinding (including reverse reproduction), press the play button 2-52, stop button 2-53, or shuttle ring 2-51 of the jog shuttle unit 2-5 (FIG. 7), respectively. It can be done by operating.

Next, by the information recording / reproducing apparatus 2-12, the pronunciation symbol information of the publication transmitted from the recording medium (for example, the center system 1 is changed to the information recording / reproducing apparatus 2-1).
FIG. 2 is a diagram showing an operation in the case of reproducing phonetic symbol information of a publication recorded in a recording medium which is recorded in advance by (2) and performing voice synthesis using this, that is, an operation in the case of reading a publication. This will be described with reference to the flowchart of No. 9.

In this case, first, when the power switch of the central control unit 2-1 (FIG. 5) is turned on, the central control unit 2-1 controls the display 2-7, whereby a predetermined menu screen is displayed. It is displayed on the display 2-7 or the central controller 2-1 controls the voice synthesizer 2-6, whereby the content of the menu screen is output from the voice synthesizer 2-6 as a synthetic voice.

Then, the user selects, from the menu, one for reading the publication. In this selection, the voice corresponding to the menu for reading the publication is selected by the voice recognition device 2-.
2 (FIG. 6) can be performed by inputting from the microphone 2-21, the keyboard 2-3, the button section 2-
4 or by operating the jog shuttle unit 2-5 (hereinafter, appropriately, the voice recognition device 2-
2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 is operated).

In the case of using the keyboard 2-3, type the character corresponding to the menu for reading aloud the publication, or move the cursor to the portion corresponding to the menu for reading aloud displayed on the display 2-7. This is done by operating a key (not shown) to make a selection. In the case of using the button section 2-4 (Fig. 7), the display 2-7
The part corresponding to the menu for reading the publication displayed in is selected by operating the left button 2-33 and the right button 2-34, and operating the Yes / confirm button 2-31 or the confirm button 2-38. And then confirm. Furthermore, in the case of using the jog shuttle unit 2-5 (Fig. 7),
The portion corresponding to the menu for reading aloud the publication displayed on the display 2-7 is selected by operating the shuttle ring 2-51 (rotating right or left), and selecting the button portion 2.
-4 Yes / Enter button 2-31 or Enter button 2
This is done by operating -38 to confirm.

The central controller 2-1 selects a menu for reading a publication as described above,
An operation mode for reading a publication (hereinafter referred to as a reading mode) is set. First, in step S2-1 shown in FIG. 9, a publication (reading text) to be read is selected. That is, the central control device 2-1 controls the information recording / reproducing device 2
-12 (the recording medium incorporated in the information recording / reproducing apparatus 2-12), the title of the publication in which the phonetic symbol information is recorded is displayed on the display 2-7, or the synthesized speech is generated from the speech synthesizer 2-6. Output by (below,
It is said to be output from the display 2-7 or the voice synthesizer 2-6).

The user selects a title desired to be read aloud from the titles output from the display 2-7 or the voice synthesizer 2-6. As with the case of selecting the menu described above, the selection of this title outputs the voice corresponding to the title of the publication to the microphone 2 of the voice recognition device 2-2.
It can be performed by inputting from -21, or by operating the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5.

When the title of the publication to be read is selected, the flow proceeds to step S2-2, and the reading form is selected.
Here, examples of the recitation mode include a mode in which recitation is performed from the beginning of the text of the publication, a mode in which recitation is started from a portion in which the recitation is interrupted last time, and a mode in which only the table of contents of the publication is reciteed.

Even when this reading mode is selected, the reading mode menu is displayed on the display 2-7 as in the case of selecting the reading text in step S2-1.
Alternatively, it is output from the voice synthesizer 2-6. Then, a desired one of them is selected by the user operating the voice recognition device 2-2, the keyboard 2-3, the button unit 2-4, or the jog shuttle unit 2-5.

After the processing of step S2-2, step S2
-3, a question menu asking (whether to ask) whether or not the reading position needs to be set is output from the display 2-7 or the voice synthesizer 2-6. Then, the user operates the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 to select whether or not to set the reading position. Then, the voice recognition device 2-2, the keyboard 2-3, the button part 2-4, or the jog shuttle part 2-5 outputs a selection signal corresponding to the operation of the user to the central control device 2-1. In the central control device 2-1, based on the selection signal,
It is determined whether or not the reading position needs to be set.

If the central controller 2-1 determines that the reading position need not be set, step S2-4
And skip to step S2-5. If it is determined in the central controller 2-1 that the reading position needs to be set, the process proceeds to step S2-4, and the reading position selection menu is displayed on the display 2-7 or the voice synthesizer 2-6. Is output. And by the user
Voice recognition device 2-2, keyboard 2-3, button section 2-
4 or the jog shuttle unit 2-5 is operated to select the reading position.

Here, at the time of selecting the reading position in step S2-4, a chapter or paragraph for which reading is desired is selected, and reading is started from the beginning of the chapter or paragraph, or a number is read from the beginning. Read only sentences, read only keywords for preset fast reading (details will be described later), or read only the vicinity of the random access point set by the user in step S2-10 described later. It is designed so that you can select the form to read.

After that, the process proceeds to steps S2-5 to S2-8 in order to select the voice quality and utterance speed of the synthetic voice for reading the publication output by the voice synthesizing device 2-6, and the voice recording / reproducing device 2-13. Selection of whether or not to record a synthetic sound (selection of necessity of recording) and selection of frequency characteristic of synthetic sound (selection of hearing loss correction) are performed.

That is, in steps S2-5 to S2-8, each selection menu is output from the display 2-7 or the voice synthesizer 2-6. Then, depending on the user, the voice recognition device 2-2, the keyboard 2-3,
The button unit 2-4 or the jog shuttle unit 2-5 is operated to select the desired voice quality and speech rate of the synthesized sound, whether or not to record the synthesized sound, and the desired frequency characteristic of the synthesized sound.

Here, the voice synthesizing device 2-6 outputs a voice quality selection menu of the synthesized voice, and the voice recognizing device 2-2 is operated (the voice recognizing device 2-2 (the microphone 2
If a low voice quality is selected (by inputting voice to -21)), for example, from the voice synthesizer 2-6,
The synthesized voice "What is your pitch?" Is output, and depending on the user,
The voice "low" is input to the voice recognition device 2-2.

Further, the voice synthesizing device 2-6 outputs a selection menu of the speaking speed of the synthetic voice, and the voice recognizing device 2-2.
When a low utterance speed is selected by operating, the voice synthesizing device 2-6 outputs a synthesized voice “What is the voice utterance?” And the user outputs the voice “slow”. It is input to the recognition device 2-2.

Further, the voice synthesizing device 2-6 outputs a selection menu as to whether or not to record the synthetic voice, and the voice recognizing device 2-2 is operated to select the recording of the synthetic voice. In this case, for example, the voice synthesizer 2-
From 6, the synthesized sound “Do you want to record?” Is output, and the user inputs the voice “Yes” to the voice recognition device 2-2.

After the processing of steps S2-5 to S2-8,
In step S2-9, the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, so as to start reading aloud,
Alternatively, the jog shuttle unit 2-5 is operated. Then,
In the central control device 2-1, the information recording / reproducing device 2-1
The phonetic symbol information of the publication recorded in No. 2 is read out and output to the voice synthesizer 2-6. As a result, in the voice synthesizer 2-6, voice synthesis is performed as described above, and a synthetic voice is output. That is, the publication selected in step S2-1 is read in the recitation mode selected in step S2-2 from the recitation position selected in step S2-4. Furthermore, at this time, the speech synthesizer 2-6
From step S2-5, S2-6, or S2-8
The synthesized voice having the selected voice quality, vocalization speed, or frequency characteristic is output.

When the reading is started by operating the voice recognition device 2-2 (by inputting voice to the voice recognition device 2-2), for example, the voice "reading" is changed to the voice recognition device. 2-2 is input. When using the keyboard 2-3, for example, the character string "roudoku"
Is typed. Further, when the button unit 2-4 is used, the reading button 2-37 (FIG. 7) is operated.

When the reading is started as described above,
If it is selected in step S2-7 to record the read-aloud synthetic sound, the voice recording / reproducing device 2-13 starts recording the read-aloud synthetic sound. Then, the process proceeds to step S2-10, and points are set for random access as needed. This point is set at the reading position where the user wants to hear the synthesized sound again, for example, at the voice recognition device 2-2 and the keyboard 2.
-3, or the button section 2-4 (in the case of the button section 2-4, the setting button 2-40 (FIG. 7) among them) is operated.

At the reading position where this point (random access point) is set, the voice recognition device 2-2,
By operating the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5, immediate access can be made.

After that, the process advances to step S2-11, and the voice recognition device 2-2 and the keyboard 2 are arranged so as to interrupt the reading.
-3 or the button portion 2-4 is determined by the user. In step S2-11, when it is determined that the voice recognition device 2-2, the keyboard 2-3, or the button unit 2-4 is operated by the user so as to interrupt the reading, the reading of the publication is combined. The sound output is stopped, and the audio recording / reproducing apparatus 2-13
If recording is in progress, the operation is stopped,
It returns to step S2-2.

That is, in this case, the reading style for reading the publication, the voice quality of the synthesized sound, the speaking speed, and the like are selected again.

In step S2-11, the voice recognition device 2-2 and the keyboard 2 are arranged so as to interrupt the reading.
-3 or the button unit 2-4 is determined not to be operated by the user, the process proceeds to step S2-12, the voice recognition device 2-2, the keyboard 2-3, so that the reading is finished. Alternatively, it is determined whether the button portion 2-4 has been operated by the user.

At step S2-12, the voice recognition device 2-2, the keyboard 2-3,
Alternatively, if it is determined that the button section 2-4 is not operated, the process returns to step S2-10. Also, step S2
In -12, when it is determined that the voice recognition device 2-2, the keyboard 2-3, or the button unit 2-4 is operated to end the reading, the output of the synthetic sound is stopped,
Further, when the audio recording / reproducing apparatus 2-13 is in the recording operation, the operation is stopped and the processing is ended.

Next, by the information recording / reproducing apparatus 2-12, the recording medium (for example, the data of the publication transmitted from the center system 1 (including character data, image data such as photographs and drawings)) Recording / reproducing apparatus 2-12
The operation of reproducing the data of the publication recorded on the recording medium which has been recorded in advance and displaying it on the display 2-7 will be described with reference to the flowchart of FIG.

In this case as well, the central controller 2-1 is first used in the same manner as in the case of reading the above-mentioned publication.
Is put into an operating mode for displaying the publication. And
In step S2-21 shown in FIG. 10, a publication (display text) to be displayed is selected. That is, the central control device 2-1 displays the title of the publication whose data is recorded on the information recording / reproducing device 2-12 (recording medium incorporated in the information recording / reproducing device 2-12) on the display 2-.
7 or the voice synthesizer 2-6.

Then, the desired title among the titles output from the display 2-7 or the voice synthesizer 2-6 is selected by the user from the voice recognizer 2-2, the keyboard 2-3, the button section 2-4, or the user. Jog shuttle part 2-
5 is operated and selected.

When the title of the publication to be displayed is selected, the process proceeds to step S2-22, and the display form is selected. Here, the display form includes, for example, a form in which the display is performed from the beginning of the text of the publication, a form in which the display is started from a portion where the previous display was interrupted, and a form in which only the contents of the publication are displayed.

In the selection of the display form, the menu of the display form is output from the display 2-7 or the voice synthesizing device 2-6, and the desired one is selected by the user from the voice recognizing device 2-2 and the keyboard. This is performed by operating and selecting 2-3, the button section 2-4, or the jog shuttle section 2-5.

After the processing of step S2-22, step S2
Proceed to 2-23, and the question menu that asks (question) whether the display position needs to be set is displayed on the display 2-7.
Alternatively, it is output from the voice synthesizer 2-6. Then, the user recognizes the voice recognition device 2-2 and the keyboard 2-.
3, the button portion 2-4, or the jog shuttle portion 2-5 is operated to select whether or not to set the display position.
Then, the voice recognition device 2-2, the keyboard 2-3, the button part 2-4, or the jog shuttle part 2-5 outputs a selection signal corresponding to the operation of the user to the central control device 2-1. The central controller 2-1 determines whether or not it is necessary to set the display position based on the selection signal.

If the central controller 2-1 determines that the setting of the display position is not necessary, step S2-2
4 is skipped, and the process proceeds to step S2-25. Also,
When the central controller 2-1 determines that the display position needs to be set, the process proceeds to step S2-24, and the display position selection menu is output from the display 2-7 or the voice synthesizer 2-6. It Then, the user operates the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 to select the display position.

Here, at the time of selecting this display position, as in the case of selecting the reading position described above, the chapter or paragraph desired to be displayed is selected and the display is started from the beginning of the chapter or paragraph. Or only a few sentences from the beginning, only preset keywords for quick reading, or only the vicinity of the random access point set by the user in step S2-26 described later. You can select the form to do.

After that, the process proceeds to step S2-25, and the voice recognition device 2-2 and the keyboard 2 are operated so as to start the display.
-3, button section 2-4, or jog shuttle section 2-5
Is operated. Then, in the central controller 2-1
The data of the publication recorded in the information recording / reproducing apparatus 2-12 is read out and output to the display 2-7 for display. That is, the publication selected in step S2-21 is displayed in the display form selected in step S2-22 from the display position selected in step S2-24.

When this display is started by operating the voice recognition device 2-2 (by inputting voice to the voice recognition device 2-2), for example, the voice "display" is displayed by the voice recognition device. 2-2 is input. In the case of using the keyboard 2-3, for example, the character string "hyouji" is typed. Further, in the case of using the button section 2-4, the display button 2-39 (FIG. 7) is operated.

When display is started as described above,
The user can scroll the display screen of the display 2-7 by operating the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, and the jog shuttle section 2-5. .

Then, the processing proceeds to step S2-26, and the points for random access are set if necessary. The settings for this random access point are
The user scrolls the display screen as described above and, for example, at the display position at which he / she wants to see it again, the voice recognition device 2-2, the keyboard 2-3, or the button section 2-
4 (in the case of the button section 2-4, as described above, it is performed by operating the setting button 2-40 (FIG. 7)).

At the display position where this random access point is set, the voice recognition device 2-2 and the keyboard 2-
By operating 3, the button section 2-4 or the jog shuttle section 2-5, it is possible to access immediately.

After that, the process proceeds to step S2-27, and the voice recognition device 2-2 and the keyboard 2 are operated so as to end the display.
-3 or the button portion 2-4 is determined by the user.

At step S2-27, the voice recognition device 2-2, the keyboard 2-3,
Alternatively, if it is determined that the button portion 2-4 is not operated, the process returns to step S2-26. Also, step S2
In -27, when it is determined that the voice recognition device 2-2, the keyboard 2-3, or the button unit 2-4 is operated so as to end the display, the display by the display 2-7 is stopped, and the process is performed. To finish.

Next, by the information recording / reproducing apparatus 2-12, the recording medium (for example, the data of the publication transmitted from the center system 1 (including character data, image data such as photographs and drawings) is converted into information). Recording / reproducing apparatus 2-12
The operation of reproducing the data of the publication recorded on the recording medium which has been recorded in advance and printing it out from the printer 2-8 will be described with reference to the flowchart of FIG.

Also in this case, the central controller 2-1 is first used in the same manner as in the case of reading the above-mentioned publication.
Is put into an operating mode to print out the publication.
Then, in step S2-31 shown in FIG. 11, a process of selecting a publication (output text) to be printed out is performed. That is, the central control device 2-1 displays the title of the publication whose data is recorded in the information recording / reproducing device 2-12 (the recording medium incorporated in the information recording / reproducing device 2-12) on the display 2-7 or the audio. Output from the synthesizer 2-6.

Then, the desired title among the titles output from the display 2-7 or the voice synthesizing device 2-6 is selected by the user from the voice recognizing device 2-2, the keyboard 2-3, the button section 2-4, or the user. Jog shuttle part 2-
5 is operated and selected.

When the title of the publication to be printed out is selected, the flow advances to step S2-32 to select the output form. Here, the output form includes, for example, a form of printing out from the beginning of the body of the publication, a form of starting printout from the part where the previous printout was interrupted, a form of printing out only the table of contents of the publication, etc. .

In the selection of the output form, the menu of the output form is output from the display 2-7 or the voice synthesizing device 2-6, and a desired one of them is selected by the user from the voice recognizing device 2-2 and the keyboard. This is performed by operating and selecting 2-3, the button section 2-4, or the jog shuttle section 2-5.

After the processing of step S2-32, step S
Proceed to 2-33, the question menu for inquiring (asking question) whether output position setting is necessary is displayed on the display 2-7.
Alternatively, it is output from the voice synthesizer 2-6. Then, the user recognizes the voice recognition device 2-2 and the keyboard 2-.
3, the button portion 2-4, or the jog shuttle portion 2-5 is operated to select whether or not to set the output position.
Then, the voice recognition device 2-2, the keyboard 2-3, the button part 2-4, or the jog shuttle part 2-5 outputs a selection signal corresponding to the operation of the user to the central control device 2-1. The central control device 2-1 determines whether or not it is necessary to set the output position based on the selection signal.

If the central controller 2-1 determines that the setting of the output position is not necessary, step S2-3
4 is skipped and the process proceeds to step S2-35. Also,
If the central controller 2-1 determines that it is necessary to set the output position, the process proceeds to step S2-34, and the output position selection menu is output from the display 2-7 or the voice synthesizer 2-6. It Then, the user operates the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 to select the output position.

Here, at the time of selecting this output position, as in the case of selecting the reading position described above, the chapter or paragraph for which printout is desired is selected, and the printout is performed from the beginning of the chapter or paragraph. Start, print only a few sentences from the beginning, print only preset keywords for quick reading, or read aloud by the user (Fig. 9)
It is possible to select a mode in which only the vicinity of the random access point set in the display process (FIG. 10) is printed out.

After that, the process proceeds to step S2-35, and the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 is operated so as to start the printout. Then, the central controller 2-1
At, the data of the publication recorded in the information recording / reproducing apparatus 2-12 is read out, outputted to the printer 2-8 and printed out. That is, the publication selected in step S2-31 is printed out in the output form selected in step S2-32 from the output position selected in step S2-24.

When the printout is started by operating the voice recognition device 2-2 (by inputting voice to the voice recognition device 2-2), for example, the voice "output" is voice recognition. It is input to the device 2-2. In the case of using the keyboard 2-3, for example, the character string "syutsuryoku" is typed. Furthermore, the button part 2
-4, the output button 2-41 (FIG. 7) is operated.

After that, when the printout is completed, the processing is completed.

The above reading, display, and printout are the information of the publications transmitted from the center system 1 when the information recording / reproducing apparatus 2-12 has a removable recording medium. It is possible to set the recording medium on which the information of the publication has already been recorded in the information recording / reproducing apparatus 2-12, instead of recording the information on the recording medium.

Further, as described above, the central control unit 2-
1 to refer to the time counted by the timer circuit 2-11, and to access the center system 1 at a time when the communication charge is low (for example, at midnight). When receiving the information of the publication searched from Fig. 1), the reading style, reading position, voice quality of synthesized speech, etc. are selected in advance,
Based on the selected value, the received publication can be read aloud by the voice synthesizer 2-6, and the synthesized voice of the read aloud can be recorded by the voice recording / playback device 2-13.

In this case, the user, for example, on the next day,
Immediately listen to the synthetic speech that read the publication.

Next, in this terminal 2, when the audio recording / reproducing device 2-13 is, for example, a DAT (digital audio tape) device, the DAT in which audio has already been recorded is recorded in the audio recording / reproducing device 2-13. Thus, it can be reproduced as shown in FIG.

That is, first, in step S2-41,
DAT is loaded into the audio recording / reproducing device 2-13, and further input from the central control device 2-1 (audio recording / reproducing device 2
-13. Command input from the central control device 2-1 is selected instead of inputting a command by operating an operation unit (not shown) equipped in -13, and steps S2-42 to S2 are performed. Go to any one of -44.

When the process proceeds to step S2-42, the left button 2-34 or right button 2-35 of the button section 2-4 (FIG. 7) or the play button 2-52 of the jog shuttle section 2-4. Is operated, the audio recording / reproducing device 2-13 is controlled so that the central control device 2-1 starts the reproduction (reverse reproduction) of the DAT.

If the process proceeds to step S2-43, the voice "reproduction" is input to the voice recognition device 2-2, so that the central controller 2-1 causes the DAT.
Recording / reproducing apparatus 2-13 so that the reproduction of
Is controlled. Further, when the process proceeds to step S2-44, the keyboard 2-3 is used to display the character string "sais
The central control unit 2-
The audio recording / reproducing device 2-13 is controlled so that the reproduction of the DAT is started by 1.

Next, fast-forwarding or rewinding the DAT loaded in the audio recording / reproducing apparatus 2-13 is performed as follows. That is, first, the input from the central controller 2-1 is selected. Then, when, for example, a voice "fast forward" or "rewind" is input to the voice recognition device 2-2, voice recording is performed so that the central control device 2-1 fast forwards or rewinds the DAT, respectively. Playback device 2-13
Is controlled.

When the shuttle ring 2-51 of the jog shuttle unit 2-5 (FIG. 7) is rotated right or left, the central controller 2-1 rotates the shuttle ring 2-51. The audio recording / reproducing device 2-13 is controlled so that the fast-forwarding or rewinding of the DAT is performed at a speed according to the angle.

Further, even when the character string "hayaokuri" or "makimodosi" is typed in with the keyboard 2-3, the central controller 2-1 activates the DAT.
The audio recording / reproducing apparatus 2-13 is controlled so that the fast-forwarding or the rewinding of is performed.

Next, DAT reproduction, fast-forwarding, or rewinding is stopped in the following manner. That is, first, the input from the central controller 2-1 is selected. Then, when, for example, a voice "stop" is input to the voice recognition device 2-2, the voice recording / reproducing device is configured so that the reproduction, fast-forwarding, or rewinding of the DAT is stopped by the central control device 2-1. 2-13 is controlled.

Also, when the No button 2-32 of the button section 2-4 (FIG. 7) or the stop button 2-51 of the jog shuttle section 2-5 (FIG. 7) is operated, the central controller 2- 1 controls the audio recording / reproducing apparatus 2-13 so that the reproduction, fast-forwarding, or rewinding of DAT is stopped.

Furthermore, even if the character string "teisi" is typed in with the keyboard 2-3, the central controller 2-1 stops the DAT reproduction, fast-forwarding or rewinding so as to stop the sound. The recording / reproducing device 2-13 is controlled.

Next, the reception of the radio program or the television program by the broadcast receiving device 2-10 is performed as follows. That is, first, the input from the central controller 2-1 is selected. Then, for example, when a voice "radio" or "television" is input to the voice recognition device 2-2,
The central control device 2-1 controls the broadcast receiving device 2-10 so as to receive the radio program or the television program, respectively.

Also, when a radio button or a television button (not shown) of the button section 2-4 (FIG. 5) is operated, the radio program or the television program is received by the central controller 2-1. Thus, the broadcast receiving device 2-10 is controlled.

Further, even when the character string "radio" or "tv" is typed in through the keyboard 2-3, the central control unit 2-1 receives the radio program or the television program, respectively. So that
The broadcast receiving device 2-10 is controlled.

When the broadcast receiving apparatus 2-10 receives a radio program, its sound is output from the speaker.
Further, when the television program is received by the broadcast receiving device 2-10, the image is displayed on the display 2-7 and the sound is output from the speaker.

The channel of the radio program or the television program to be received by the broadcast receiving device 2-10 can be selected by operating the shuttle ring 2-51 of the jog shuttle section 2-5. ing.

Further, the reception of the radio program or the television program by the broadcast receiving device 2-10 can be stopped in the same manner as the case where the reproduction, the fast forward or the rewind of the DAT described above is stopped. Has been done.

As described above, the operation of the terminal 2 can be controlled not only by the operation of the keyboard 2-3 but also by the voice input and the operation of the button section 2-4 or the jog shuttle section 2-5. In addition to being performed, information can be output by both voice and display, so that even a user with a physical disability can easily use it.

Next, FIG. 13 is a block diagram showing a detailed structure of the voice input type database retrieval / transmission system 3 of the center system 1 shown in FIG. Communication control unit 3-
A communication line (a telephone line in this embodiment as described above) connected to the terminal 2 is connected to the terminal 1.
Then, the communication control section 3-1 is configured to perform transmission / reception control of information with the terminal 2 based on a predetermined communication standard. Further, a plurality of communication lines, that is, a plurality of terminals can be connected to the communication control section 3-1. In this case, the communication control section 3-1
A plurality of transactions are processed in parallel by a time division method, for example.

Note that, in processing a plurality of transactions, in addition to the time sharing as described above, FIG.
It is also possible to provide a plurality of voice input type database retrieval / transmission systems 3 shown in FIG.

Further, the communication control unit 3-1 determines from the received data that the terminal attributes of the terminal 2 (specification of hardware (for example, the terminal 2 is equipped with a voice synthesizer, the terminal dedicated to the visually impaired). When transmitting the information (signal) to the terminal 2, the information is converted into a format corresponding to the terminal attribute and transmitted.

The communication control section 3-1 may be configured to allow the terminal 2 to perform the format conversion process without performing the format conversion process corresponding to the terminal attribute as described above.

Hereinafter, for simplification of explanation, it is assumed that one terminal 2 is connected to the communication control section 3-1.

First, when the communication control unit 3-1 receives the data, the data is the data of the voice itself or other data (for example, the acoustic parameter of the voice,
Command etc. (Hereinafter, referred to as non-voice data), and the determination signal is output to the system control unit 3-3.

The system control section 3-3 has a communication control section 3-
When the determination signal is received from 1, the switch control of the switch 3-2 is performed based on the determination signal. That is, when the determination signal indicates that the received data is the data of the voice itself, the system control unit 3-3
The switch 3-2 is caused to select the terminal a side. If the determination signal indicates that the received data is non-voice data, the system control unit 3-3 causes the switch 3-2 to select the terminal A side.

In the communication control section 3-1, since the communication line connected thereto is a telephone line as described above, there is a case where a telephone call is accidentally made from the telephone and the voice itself is received. is there. In this case, the communication control unit 3-
From 1, a determination signal indicating that the received data is voice itself is output to the system control unit 3-3,
As a result, in the system control unit 3-3, the switch 3-
2 is switched to the terminal a side. Furthermore, in this case, the system control unit 3-3 switches the switch 3-9 to the terminal b side, whereby the voice synthesis unit 3-7 causes the switch 3-9 to operate.
It is connected to the communication control unit 3-1 via -9 and 3-2.

When the switch 3-9 is switched to the terminal b side, the voice synthesizing unit 3-7 is controlled by the system control unit 3-3 and stored in the voice synthesizing database 3-8 in advance. For example, the voice data is read out, and based on the voice synthesis data, a synthetic voice having contents such as an incorrect telephone number is generated.

This synthesized sound is generated by switches 3-9 and 3-
2, and is transmitted to the called telephone via the communication control unit 3-1 so that the other party who is calling is notified that the telephone number is incorrect.

On the other hand, when the data from the terminal which should be originally connected to the center system 1 such as the terminal 2, that is, the non-voice data is received, the switch 3-2 is switched to the terminal A side as described above. As a result, the non-voice data is received by the system control unit 3-3 via the communication control unit 3-1 and the switch 3-2.

Here, in the system control unit 3-3, even if the determination signal indicating that the received data is the non-voice data is received from the communication control unit 3-1, it is requested by the terminal 2. , The switch 3-2 can be switched to the terminal a side. In this case, the system control unit 3-3 switches the switch 3-2 to the terminal a side and the switch 3-9 to the terminal a side, whereby the operator connected to the terminal a. Terminal operated by (not shown)
And the terminal 2 can communicate with each other.

In this case, the user of the terminal 2 can directly inquire the operator, for example, by voice, and thus, for example, a beginner who is unfamiliar with the operation method of the terminal 2 or an unexpected situation of the user. Being able to deal with.

In addition to the above, the system control unit 3-3
However, when an unexpected situation occurs in the center system 1, such as when communication with the terminal 2 cannot be performed via the communication control unit 3-1, the switch 3-2 or 3-
9 is switched to the terminal a or a side, respectively, so that the user of the terminal 2 and the operator can directly communicate with each other, for example, by voice.

When the non-voice data is received via the switch 3-2 which selects the terminal A side, the system control section 3-3 executes a predetermined process corresponding to the non-voice data. Is supposed to do.

That is, when the non-speech data is, for example, an acoustic parameter for which speech recognition is to be performed and a command for instructing that speech recognition, the system control unit 3-3 recognizes this acoustic parameter. The recognition unit 3-4 is controlled so as to perform a voice recognition process while supplying the recognition unit 3-4. If the non-voice data is, for example, a command or other information for retrieving publication data from the book database 5 (FIG. 1), the retrieval unit 3-6 retrieves the publication data. Along with the search results,
It transmits to the terminal 2 via the switch 3-2 and the communication control unit 3-1.

Here, in the present embodiment, as described above, the central controller 2-1 (analyzer 2-
The acoustic parameters from 23) are transmitted to the center system 1. Therefore, in this case, the signal deterioration can be reduced and the communication amount can be further reduced, as compared with the case where the voice itself is transmitted through the communication line.

Further, voice recognition is performed using this acoustic parameter. In this case, the recognition performance can be improved as compared with the case of recognizing the voice itself transmitted through the communication line ( In this embodiment, a telephone line is assumed as the communication line, and for example, the communication line is PS.
TN, and when voice is transmitted via this PSTN, the voice is band-limited to a range up to approximately 3.4 kHz).

In order to configure the terminal 2 in a small size, for example, the case where the terminal 2 is configured without providing the voice recognition device 2-2 (FIG. 5) corresponds to the voice itself from the terminal 2. Even if it is transmitted, the recognition unit 3-4 can perform the voice recognition.

The recognition unit 3-4 performs a voice recognition process under the control of the system control unit 3-3, using the acoustic parameters similarly output from the system control unit 3-3. That is, the recognizing unit 3-4 vector-quantizes the acoustic parameter from the system control unit 3-3 using, for example, a code book created in advance, and stores the acoustic parameter in the recognizing database 3-5. For example, discrete H
The phoneme-based models created based on the MM method are connected in accordance with the words described in the word dictionary also stored in the recognition database 3-5. Then, this connection model calculates the probability of outputting the vector-quantized acoustic parameter sequence, and outputs the recognition result based on this probability.

Note that the speech recognition algorithm is a discrete H
The method is not limited to the MM method, for example, a continuous mixed HMM.
Any method such as a method or a method using a neural network can be used.

The recognizing unit 3-4 is also designed to be capable of recognizing speech by word recognition as described above and speech recognition by dictation (details will be described later).

Further, the recognition unit 3-4 is configured to have a wider recognition vocabulary range and higher recognition accuracy than the voice recognition device 2-2 provided in the terminal 2.

Next, the operation will be described with reference to the flowchart in FIG. First, step S3-
1, the line connection between the terminal 2 (FIG. 1) and the center system 1 is established. This is because the terminal 2 side sends the user ID and the personal identification number, and the center system 1 confirms the information.
Follow the general connection procedure when connecting to the center system.

When the line connection is made, step S3-2
In step S, the menu of services that can be provided by the center system 1 is transmitted to the terminal 2, and step S
Proceeding to 3-3, the user of the terminal 2 selects the desired one from the service menu. That is,
As described above, in the terminal 2, the service menu is output from the display 2-7 or the voice synthesizer 2-6, and a desired one of them is the voice recognizer 2-2, the keyboard 2-3, the button section 2-. 4 or the jog shuttle unit 2-5 is selected by being operated by the user.

Output of service menu (enumeration)
Can be omitted according to the predetermined information preset in the terminal 2 or the center system 1. That is, for example, if the user knows the content of the service in advance, the output of the service menu can be omitted. Further, when the user uses only a predetermined service among the services provided by the center system 1, the output of the service menu and the processes of steps S3-3 and S3-4 described later are omitted, and step S3 is performed. It is possible to provide a predetermined service immediately after the processing of -1.

When the service menu is selected, a selection signal (command) indicating the selected service menu is transmitted from the terminal 2 to the center system 1 in step S3-3. And step S3-4
Go to the center system 1 (system control unit 3-3)
Then, it is determined what service menu the selection signal (command) indicates, and the process proceeds to any of steps S3-5 to S3-8 based on the determination result.

At step S3-5, the system controller 3
In -3, a program related to newspaper service is started and a service related to newspaper is provided. In step S3-6, the system control unit 3-3 activates the program related to the magazine service to provide the service related to the magazine. Further, in step S3-7, the system control unit 3-3 activates the program related to the book search and provides the service related to the book search. In step S3-8, the system control unit 3-3 activates the program related to the new publication service and provides the service related to the new publication.

As described above, when the program for providing each service is activated, the system control section 3-3 receives the command subsequently transmitted from the terminal 2. Then, as shown in the flowchart of FIG. 15, first, in step S3-11, the received command is analyzed, and based on the analysis result, the process proceeds to one of steps S3-12 to S3-14.

In step S3-12, the system control unit 3-3 controls the recognition unit 3-4 for word recognition or phoneme recognition (voice recognition by dictation). Further, in step S3-13, the system control unit 3-3 performs the control process of the information retrieval from the book database 5 in the retrieval unit 3-6.
Furthermore, in step S3-14, the system control unit 3-
3, the transfer control processing of the data requested by the terminal 2 is performed.

Next, the details of the recognition unit 3-4 shown in FIG. 13 will be described. FIG. 16 is a block diagram showing detailed configurations of the recognition unit 3-4 and the recognition database 3-5. In the recognition database 3-5, the model and the word dictionary are classified into predetermined vocabulary groups 1, 2,
..., N, and the data are classified and stored in N groups.

Here, the category means that the word means the title of the publication, the attribute of the publication such as the author, the publisher, the date of publication, the classification, or the system control unit 3-3. It means a unit for classifying words, so to speak. In the recognition database 3-5, models and word dictionaries are grouped into those that show publication titles, attributes of publications such as author, publisher, publication date, and classification, and those that show commands. It is stored separately.

The recognition unit 3-4 includes a recognition method selection unit 3-1.
1, word recognition unit 3-12, dictation unit 3-1
3 and an attribute list creation unit 3-14.
From the system control unit 3-3 to the selection recognition unit 3-11,
A control signal is supplied together with the acoustic parameter, in which the acoustic parameter is transmitted along with the control signal in accordance with the control signal.
2 and the dictation unit 3-13.

As shown in FIG. 17, the word recognition unit 3-12 refers to the model and the word dictionary read by the selection target vocabulary selector 3-22 from the recognition database 3-5 to perform voice recognition (word recognition). Word recognizer 3 for recognition
-21, a selection target vocabulary selector 3-22 for selecting one of the vocabulary groups 1 to N from the recognition database 3-5 and reading out a model and a word dictionary belonging to the selected vocabulary group, and a recognition method selection Part 3-11
Based on the control signal from (FIG. 16), the word recognizer 3-
21 and a word recognition controller 3-23 for controlling the recognition target vocabulary selector 3-22, and operates according to the flowchart shown in FIG.

That is, in the word recognition section 3-12, first, in step S3-21, the word recognition controller 3-23 causes the recognition method selection section 3-11 (system control section 3-12).
It is determined whether the control signal from 3) is received.
When it is determined in step S3-21 that the control signal is not received, the process returns to step S3-21.

If it is determined in step S3-21 that the control signal is received, then step S3-2
2, the word recognition controller 3-23 indicates that the control signal selects any one of the vocabulary groups 1 to N of the recognition database 3-5 (hereinafter referred to as selecting a category). It is determined whether or not it is a thing.

The output of the control signal indicating the category from the system control section 3-3 will be described later.

If it is determined in step S3-22 that the control signal indicates selection of a category, the process advances to step S3-23, and the word recognition controller 3-23 causes the recognition target vocabulary selector 3 to be selected. -23 is controlled, whereby the recognition target vocabulary selector 3-23 causes
A model and a word dictionary belonging to the category (any of vocabulary groups 1 to N) indicated by the control signal are read out,
The process ends.

On the other hand, when it is determined in step S3-22 that the control signal does not indicate that the category is selected, the process proceeds to step S3-24, and the control signal is recognized by the word recognizer 3-21. It is determined whether or not the processing is instructed. When it is determined in step S3-24 that the control signal does not instruct the word recognizer 3-21 to perform the recognition process, the process proceeds to step S3-25.
An error signal indicating that the control signal cannot be decoded is obtained as a recognition result from the word recognition controller 3-23 via the attribute list creation unit 3-14 (FIG. 16) and the system control unit 3-3.
Is output to and the processing ends.

If it is determined in step S3-24 that the control signal instructs the word recognizer 3-21 to perform the recognition process, the process proceeds to step S3-26.
The recognition database 3-by the word recognizer 3-21.
5, the model and the word dictionary read by the selection target vocabulary selector 3-22 are referred to, and the acoustic parameters supplied from the recognition method selection unit 3-11 (FIG. 16) are used to perform the discrete HMM method. Speech recognition processing is performed (for example, the Viterbi algorithm calculates the probability that an acoustic parameter sequence is observed from a model (word model) in which phoneme-based models are connected).

The word obtained as a result and its probability (hereinafter referred to as score) are output as the recognition result,
The process ends.

Next, FIG. 19 shows the detailed structure of the dictation unit 3-13 shown in FIG. The dictation type recognition unit 3-31 is a recognition method selection unit 3-11.
Voice recognition (dictation) is performed by performing Viterbi beam search for an HMM phoneme model continued by a bigram or the like using the supplied acoustic parameters or performing template matching continuously.
And output the result (dictation result).

Incidentally, the dictation type recognition unit 3-31
Is a speech recognition based on the dictation conditions (to be described later) stored in the dictation condition storage unit 3-32 and the past dictation results stored in the recognition result storage unit 3-33 (referred to as necessary). Is designed to do.

The dictation condition storage unit 3-22 is
As will be described later, the correction condition (dictation condition) of the dictation result pointed out by the user of the terminal 2 is stored.

Here, the correction condition (dictation condition) is, for example, when a dictation result in which the first character is “A” is output with respect to a certain input voice (acoustic parameter), the dictation result is output to the terminal 2. If it is transmitted and the user of the terminal 2 points out that the first character of the dictation result is incorrect, do not make the first character "A" when dictating the input voice again. Means a condition (constraint).

The correction condition (dictation condition) is used when the user of the terminal 2 points out an error in the dictation result.
-3 is supplied as a control signal.

In the recognition result storage unit 3-33, the dictation type recognition unit 3-31 clears the stored contents immediately before the dictation for a certain input voice is performed for the first time, and then the dictation type recognition unit 3-33. The dictation result output from 31 is stored. This recognition result storage unit 3-33
If the past dictation result is stored in the dictation type recognition unit 3-31, the dictation type recognition unit 3-31 does not make the result the same as the past dictation result, that is, the same erroneous dictation for the same voice. Dictation is performed so that the result is not output.

Next, the dictation unit 3-13 and the system control unit 3-3 for controlling it (FIG. 13).
The operation of will be described. In the system control unit 3-3,
First, step S3-31 shown in the flowchart of FIG.
In step 3, a control signal is output to the dictation unit 3-13 (FIG. 19) so as to clear the stored contents (dictation condition) of the dictation condition storage unit 3-32, the process proceeds to step S3-32, and the sound from the terminal 2 is output. The parameter is received, and the process proceeds to step S3-33. In step S3-33, a control signal for instructing voice recognition using the acoustic parameter is output to the dictation unit 3-13 together with the acoustic parameter from the terminal 2.

After that, when the dictation result (voice recognition result) is transmitted from the dictation unit 3-13, the process proceeds to step S3-34, and it is determined whether or not the dictation result is correct. Here, this determination is performed as follows. That is, first, the dictation result is transmitted to the terminal 2, and the display 2-7 or the voice synthesizer 2-6 is displayed at the terminal 2.
Is output from. Depending on the user of the terminal 2, it is pointed out whether this dictation result is correct or incorrect (speech recognition device 2-2, keyboard 2-
3, which is pointed out by operating the button section 2-4 or the jog shuttle section 2-5), whereby the terminal 2 transmits a determination signal indicating whether or not the dictation result is correct. System control unit 3-3
Then, based on this determination signal, it is determined whether or not the dictation result is correct.

If it is determined in step S3-34 that the dictation result is incorrect, that is, it is incorrect, the process proceeds to step S3-36, and the same voice is uttered again to the terminal 2 or the dictation result is confirmed. A message is sent to point out what is wrong or correct (enter dictation conditions).

Therefore, after this, the user operates the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5, whereby the dictation this time from the terminal 2. The acoustic parameter of the same voice as that of the voice or the dictation condition is transmitted.

In this case, in addition to the acoustic parameter or dictation condition, the dictation result corrected to the correct one may be transmitted from the terminal 2 as described later, and the system control unit 3 -3, when the corrected correct dictation result is received, interrupts the processing from step S3-37 and ends the processing.

After the processing of step S3-36, step S
Proceeding to 3-37, it is determined whether or not the signal transmitted from the terminal 2 is an acoustic parameter. When it is determined in step S3-37 that the signal transmitted from the terminal 2 is an acoustic parameter, the process returns to step S3-33 to perform speech recognition using the acoustic parameter from the terminal 2 and the acoustic parameter. The control signal instructing to perform is output to the dictation unit 3-13 again, and the processing from step S3-34 onward is repeated.

If it is determined in step S3-37 that the signal transmitted from the terminal 2 is not the acoustic parameter, that is, the dictation condition, the process proceeds to step S3-38, and the dictation condition is the control signal. Is output to the dictation unit 3-13, and the process returns to step S3-33.

On the other hand, if it is determined in step S3-34 that the dictation result is correct, the flow advances to step S3-35 to output the dictation result to the attribute list creation unit 3-14 (FIG. 16) described later. The instructed control signal is output to the dictation unit 3-13, and the process ends.

Next, the dictation unit 3-13 follows the control of the system control unit 3-3 as described above.
It operates as shown in the flowchart of FIG. That is,
In step S3-41, the dictation unit 3-13 (FIG. 19) recognizes the dictation type recognition unit 3-3.
1 determines whether or not the control signal from the system control unit 3-3 is received.

If it is determined in step S3-41 that the control signal has not been received, step S3-41
Return to. If it is determined in step S3-41 that the control signal is received, the process proceeds to step S3-42, and the control signal is stored in the dictation condition storage unit 3-32 by the dictation type recognition unit 3-31. Is instructed to clear (indicating an initialization request). When it is determined in step S3-42 that the received control signal indicates an initialization request, the process proceeds to step S3-43, the storage contents of the dictation condition storage unit 3-32 are cleared, and the process ends. To do. In step S3-43, when the stored contents of the dictation condition storage unit 3-32 are cleared, the stored contents of the recognition result storage unit 3-33 are also cleared at the same time.

On the other hand, when it is determined in step S3-42 that the received control signal does not indicate the initialization request, the process proceeds to step S3-44, and the control signal is under the dictation condition (dictation correction condition). It is determined whether or not there is. When it is determined in step S3-44 that the control signal is the dictation condition, the process proceeds to step S3-50, and the dictation condition is the dictation type recognition unit 3-31.
Is output from the dictation condition storage unit 3-32 and stored therein.

Here, if the control signal is a dictation condition, the dictation type recognition unit 3-3 has already been used.
In No. 1, dictation for a certain voice is performed at least once.

Therefore, when the control signal which is the dictation condition is received, the dictation condition is stored in the dictation condition storage unit 3-32 by the dictation type recognition unit 3-31 after storing the dictation condition. The stored dictation condition is referred to, and dictation on the same voice is performed again. Then, when the dictation result is output from the dictation type recognition unit 3-31, step S3-
Proceed to 48.

On the other hand, if it is determined in step S3-44 that the control signal is not in the dictation condition, the flow advances to step S3-45 to determine whether or not the control signal is a command instructing voice recognition. It When it is determined in step S3-45 that the control signal is a command instructing voice recognition, step S3
Proceeding to -47, the dictation type recognition unit 3-31 performs dictation using the control signal and the acoustic parameter supplied from the system control unit 3-3.

After that, the dictation type recognition unit 3-3
When the dictation result is output from 1, the process proceeds to step S3-48, the dictation result is stored in the recognition result storage unit 3-33, and the process proceeds to step S3-49. In step S3-49, the recognition result storage unit 3-
The dictation result stored in 33 is output to the system control unit 3-3 (FIG. 13), and the process ends.

On the other hand, if it is determined in step S3-45 that the control signal is not the command for instructing voice recognition, the process proceeds to step S3-46, and the recognition result storage section 3
The latest dictation result stored in −33 is output to the attribute list creation unit 3-46 as the final dictation result (correct dictation result), and the process ends.

Returning to FIG. 16, the attribute list creating section 3-14.
Is a word recognition unit 3-12 or a dictation unit 3-12.
Recognition result output from 13 (word and its likelihood)
Is combined with a category supplied as a control signal from the system control unit 3-3 (hereinafter, this combination is referred to as an attribute list), and is output to the system control unit 3-3 as a final voice recognition result.

Thereafter, the system control unit 3-3 refers to the attribute list as the voice recognition result output from the recognition unit 3-4, and performs a predetermined process.

Here, the system control unit 3-3 causes the recognition unit 3-4 (word recognition unit 3-12 or dictation unit 3-13) to recognize the voice (acoustic parameter).
In most cases is a voice as a reply transmitted from the terminal 2 in response to an inquiry (question) from the system control unit 3-3. Therefore, the system control unit 3
-3, a voice (word) to be recognized by the recognition unit 3-4
You can predict the category to which you belong.

Therefore, when the recognition unit 3-4 performs voice recognition, the system control unit 3-3 belongs to the category to which the voice to be recognized belongs (as described above, the voice is the title of the publication or the author. Or command etc.)
Is output to the recognition unit 3-4 as a control signal.

Since the word recognition unit 3-12 outputs the word and its likelihood as the voice recognition result, the attribute list creating unit 3-14 selects from the voice recognition result as follows.
For example, a word having a likelihood equal to or higher than a predetermined threshold is detected, the detection result (word and its likelihood), and the system control unit 3-3.
From the category supplied as a control signal from (eg, category, (word, likelihood), (word, likelihood), ...
., (Word, likelihood)) is output as an attribute list in a format such as. However, the word is
For example, they are arranged in descending order of likelihood. In the word recognition unit 3-12, the likelihood is normalized such that its value is output in the range of 0 to 1, for example, and the larger the value, the higher the likelihood. It

Furthermore, since one word (character string) is output from the dictation unit 3-13 as a voice recognition result, the attribute list creating unit 3-14 uses the word (character string) and the system control unit. The category supplied as a control signal from 3-3 is output as an attribute list in a format such as (category, (word, likelihood)). However, in this case, the likelihood is 1 which is the largest value.

Next, step S3-3 of FIG. 20 described above.
At 6, after a message is sent to the terminal 2 again to say the same voice or to point out the incorrect or correct part of the dictation result (enter the dictation condition),
For example, the operation of the terminal 2 (central control unit 2-1) (FIG. 5) when the user of the terminal 2 operates the button section 2-4 and the jog shuttle section 2-5 to correct the dictation result. 22 will be described with reference to the flowchart of FIG.

First, as described above, when it is determined in step S3-34 in FIG. 20 whether or not the dictation result is correct, the dictation result is transmitted to the terminal 2, so the terminal 2 (central control Device 2
-1) (FIG. 5), the dictation result is displayed on the display 2-7 in step S3-61, the process proceeds to step S3-62, and the jog shuttle unit 7 is operated.
It is determined whether the shuttle ring 2-51 (Fig. 7) has been rotated rightward by the user.

If it is determined in step S3-62 that the shuttle ring 2-51 has been rotated rightward by the user, the process proceeds to step S3-63, and if there is a character to be changed, which will be described later, the shuttle ring 2-51. 51
The characters to be changed are re-displayed in lexicographical order (A, I, U, E, O, K, K, ... (in order of AIUEO)) at a speed according to the rotation angle of. , Step S3-6
Return to 2.

[0285] Here, when the user utters "Asahi Shinbun", for example, in Terminal 2, an erroneous dictation result of "Asaisanbun" is received, and this is displayed on the display 2-7. The situation is shown in FIG.

As shown in FIG. 23A, the character to be changed means one character immediately after the fixed area, that is, one character immediately before the bar cursor, and the fixed area means the display 2-7. Of the character string "Asaishinbun" as the dictation result displayed in, it means the area of the character that does not require modification, and the character two characters before the position where the bar cursor is located from the beginning of the character string. Up to are defined areas.

In the terminal 2, the button section 2-4
When the left button 2-33 or the right button 2-34 in (Fig. 7) is operated once, the bar cursor moves one character to the left or right, thereby decreasing the number of characters in the defined area by one character. Alternatively, the number of characters to be changed can be set to be the character to the left or right of the character that is currently being changed. If the bar cursor is before the first character of the character string “Asaishinbun” as the dictation result, there is neither a change target character nor a fixed area.

[0288] Further, in the terminal 2, the button section 2-
4 (FIG. 7) Yes / confirm button 2-31 or confirm button 2-38 is operated, the display 2 at that time
The character string displayed at -7 is transmitted to the center system 1 (system control unit 3-3) as the correction result of the dictation result (correct dictation result).

When the shuttle ring 2-51 of the jog shuttle unit 7 (FIG. 7) is rotated right or left in the terminal 2, the character to be changed is displayed in the lexicographic order or It is designed to be displayed again in reverse lexicographic order (reverse order of AIUEO).

Specifically, as shown in FIG. 23A, for example, in the character string "Asaishinbun", if the third character "i" from the beginning is the character to be changed, the shuttle ring 2 When −51 is rotated to the right or left, the change target character “i” is changed to “u” → “e” → “o” → ...
・ Or, it is redisplayed (changed) like "A" → "N" → "Wa" → ....

Returning to FIG. 22, when it is determined in step S3-62 that the shuttle ring 2-51 is not rotated right by the user, step S3-.
Proceeding to 64, it is determined whether the shuttle ring 2-51 has been rotated left by the user. Step S
In 3-64, when it is determined that the shuttle ring 2-51 has been rotated leftward, the process proceeds to step S3-65, and when there is a character to be changed, the shuttle ring 2
The character to be changed is displayed again in the reverse order of the lexicographical order at a speed corresponding to the rotation angle of -51, and the process returns to step S3-62.

If it is determined at step S3-64 that the shuttle ring 2-51 has not been rotated left by the user, then the processing advances to step S3-66, at which the right button of the button portion (FIG. 7) is pressed. It is determined whether 2-34 has been operated by the user. Step S3-
When it is determined in 66 that the right button 2-34 has been operated, the process proceeds to step S3-67, and the confirmation area (see FIG.
3 (a)) is spread one character to the right,
The bar cursor and the character to be changed are moved to the right by one character, and the process returns to step S3-62.

On the other hand, if it is determined in step S3-66 that the right button 2-34 has not been operated, then the flow proceeds to step S3-68, where the left button 2-33 of the button section (FIG. 7) is the user. It is determined whether or not it has been operated by. In step S3-68, left button 2-33
If it is determined that is operated, the process proceeds to step S3-69, the confirmed area (FIG. 23A) is narrowed to the left by one character, and the bar cursor and the character to be changed are moved to the left by one character. It is moved and returns to step S3-62.

If it is determined in step S3-68 that the left button 2-33 has not been operated, the flow advances to step S3-70 to move the Y button on the button section 2-4 (FIG. 7).
es / confirm button 2-31 or confirm button 2-38
Is determined by the user. In Step S3-70, Yes / Enter button 2-3
If it is determined that the 1 or the confirm button 2-38 is not operated, the process proceeds to step S3-71, the input is ignored, and the process returns to step S3-62.

On the other hand, in step S3-70, Ye
When it is determined that the s / confirm button 2-31 or the confirm button 2-38 has been operated, the process proceeds to step S3-72, and the character string currently displayed on the display 2-7 is the correction result of the dictation result (correct. The dictation result) is transmitted to the center system 1 (system control unit 3-3), and the process ends.

It should be noted that the correction of the character string to be changed is not limited to the operation of the jog shuttle unit 2-5, but also the voice recognition unit 2
-2 (FIG. 5) can be caused to be performed by the central control unit 2-1 by inputting a voice. However,
In this case, it may be confused with the rewording of the entire utterance content, and therefore it is possible to correct the changed character string by voice input only when the number of characters in the re-speech is several characters (for example, one or two characters). It can be so.

Further, the dictation result and the character string in the middle of correction thereof are not displayed on the display 2-7 as shown in FIG. 23A, but are output from the voice synthesizing device 2-6 as a synthetic sound. You can In this case, from the speech synthesizer 2-6, FIG.
As shown in (b), for example, a pose is inserted before and after the character to be changed, so that even a visually impaired user can identify and correct the character to be changed. It is done like this.

Next, the user who accesses the center system 1 from the terminal 2 operates the terminal 2 (FIG. 5).
When the voice is input to the voice recognition device 2-2 of FIG.
7. The system control unit 3 when the book search service in 7 is selected and the book search is performed in the center system 1.
The operation of No. 3 will be described with reference to the flowcharts of FIGS. 24 and 25.

In this case, first, in step S3-3 of FIG. 14 described above, the user utters, for example, "search" as a voice for selecting the book search service, which is the voice recognition device 2-2 ( The acoustic parameter corresponding to the voice “search” is analyzed by the analysis unit 2-23 (FIG. 6) and transmitted from the terminal 2 to the center system 2.

Then, in step S3-81, the system control unit 3-3 (FIG. 13) determines whether or not an acoustic parameter has been received from the terminal 2. When it is determined in step S3-81 that the acoustic parameter has not been received, the process returns to step S3-81. If it is determined in step S3-81 that the acoustic parameter has been received, the process proceeds to step S3-82,
The system controller 3-3 supplies the recognition unit 3-4 with acoustic parameters from the terminal 2 and a control signal for instructing the voice recognition process.

Here, the word recognition unit 3-12 and the dictation unit 3-1 which form the recognition unit 3-4 (FIG. 16).
Which of the three is used for voice recognition is basically
It is determined by the state of the recognition method changeover switch 2-36 of the button section (FIG. 7) in the terminal 2. The recognition method changeover switch 2-36 is, for example, LE.
The LED 2-35 is provided with a D2-35, and the LED 2-35 is repeatedly turned on and off each time the changeover switch 2-36 is operated.

The LED 2 of the changeover switch 2-36
When -35 is lit, for example, the word recognition unit 3
A control signal for instructing voice recognition at -12, and a control signal for instructing voice recognition at the dictation unit 3-13 when the LED 2-35 is off, for example, are output from the terminal 2. It is transmitted to the center system 1.

If the voice corresponding to the acoustic parameter transmitted from the terminal 2 corresponds to the command (belongs to the category of the command), regardless of the state of the changeover switch 2-36 ( LED2
Regardless of whether -35 is on or off, the system control unit 3-3 causes the word recognition unit 3-12 to perform a voice recognition process (system control unit 3-3). As described above, the category to which the voice (word) to be recognized by the recognition unit 3-4 belongs can be predicted).

Therefore, in step S3-82,
The system control unit 3-3 controls the word recognition unit 3-12 of the recognition unit 3-4 (FIG. 16) to perform voice recognition using the acoustic parameter from the terminal 2, whereby the word recognition unit 3-3. At -12, voice recognition processing is performed.

Then, in step S3-83, the system control section 3-3 receives the voice recognition result (in this case, the above-mentioned attribute list) from the recognition section 3-4 (word recognition section 3-12). Whether or not it is determined. When it is determined in step S3-83 that the voice recognition result has not been received, the process returns to step S3-83. When it is determined in step S3-83 that the voice recognition result is received, the process proceeds to step S3-84, and the system control unit 3-3 causes the voice recognition result (in this case, for example, the highest likelihood). It is determined whether the voice recognition result (word) is the word "search". Step S3-
When it is determined in 84 that the voice recognition result is not the word "search", the process proceeds to step S3-85, the system control unit 3-3 performs a process corresponding to the voice recognition result, and the process ends.

On the other hand, when it is determined in step S3-84 that the voice recognition result is the word "search", the process proceeds to step S3-86, and thereafter, the process related to the book search service is performed (system control unit 3- 3 is the search mode).

That is, in step S3-86, the attribute name (category name) of the publication (as described above, the title of the publication, the author's name, the publisher, etc.) is obtained in order to obtain the keyword for performing the search. Output to terminal 2.

As a result, at the terminal 2, the attribute name of the publication is output from the display 2-7 or the voice synthesizer 2-6. Specifically, the attribute name of the publication is output so as to prompt the user to input the keyword, such as "What is the title?", "What is the author?", "What is the publisher?".

Then, the user recognizes the voice recognition device 2-2,
When a keyword is input by operating the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5, the keyword is transmitted from the terminal 2. Here, as described above, consider the case where a voice is input to the voice recognition device 2-2, that is, a keyword is uttered by the voice and the acoustic parameter thereof is transmitted from the terminal 2.

In the system control section 3-3, the terminal 2
When the acoustic parameter corresponding to the keyword from is received, the process proceeds to step S3-87, and it is determined whether or not there are still keywords to be requested to the user of the terminal 2. If it is determined in step S3-87 that there are still keywords to be requested to the user of the terminal 2, the process returns to step S3-86, and attributes of the publications of the category to which the keyword to be requested to the user of the terminal 2 belongs. First name is output.

Here, if the user does not know the attribute name (for example, author name) of the publication output from the display 2-7 or the voice synthesizer 2-6, he or she speaks, for example, "skip". By doing so, it is not necessary to input an unknown keyword.

If it is determined in step S3-87 that there are no keywords to be requested to the user of the terminal 2, the process proceeds to step S3-88 and the terminal 2 is operated.
The acoustic parameters corresponding to the keywords transmitted from are sequentially output to the recognition unit 3-4, and the recognition unit 3-4
Control is performed to sequentially perform the voice recognition processing using the acoustic parameters. Thereby, the recognition unit 3-4 performs the voice recognition process and outputs the voice recognition result in the form of the attribute list as described above.

In this case, the voice to be recognized by the recognition unit 3-4 does not correspond to the command.
The word recognition unit 3-1 corresponding to the state of the recognition method changeover switch 2-36 of the button unit (FIG. 7) in the terminal 2
2 or the dictation unit 3-13 performs voice recognition. In the following, a recognition database 3-5 in which models and word dictionaries are stored for each category (see FIG. 16).
Also, the word recognition unit 3-12 using FIG. 17) basically performs voice recognition.

After the processing in step S3-88, step S
3-89, the recognition unit 3-4 receives the voice recognition results (attribute list) of all the keywords received from the terminal 2.
It is determined whether it is received from. Step S3-89
If it is determined that the speech recognition results of all the keywords received from the terminal 2 have not been received from the recognition unit 3-4, the process returns to step S3-89. If it is determined in step S3-89 that the speech recognition results of all the keywords received from the terminal 2 have been received from the recognition unit 3-4, the process proceeds to step S3-90, and a plurality of keywords received from the terminal 2 are detected. In some cases, the voice recognition result (attribute list) corresponding to each keyword is output to the search unit 3-6 (FIG. 13).

When the voice recognition result is the above-mentioned "skip", this is not counted in the number of voice recognition results. The voice recognition result “skip” is not output to the search unit 3-6.

In step S3-90, after the voice recognition result (attribute list) is output to the search unit 3-6, a plurality of keywords (voice recognition result) are further input to the search unit 3-6.
Control is performed to bring the possible combinations together.

As a result, in the search unit 3-6, the system control unit 3-3 (the recognition unit 3 via the system control unit 3-3
-4) A plurality of keywords (voice recognition results) output from (4) are collected into possible combinations.

That is, the voice recognition is performed by the recognition unit 3-4 (see FIG. 1).
In the case of 6) in the word recognition unit 3-12, as described above, a word (recognition result) having a likelihood equal to or higher than a predetermined value is output. For example, when acoustic parameters corresponding to the attributes “title” and “author” of a publication are transmitted from the terminal 2 as a plurality of keywords, the attribute “title” is transmitted.
On the other hand, the speech recognition results “linear algebra and its application”, “I am a cat”, and so on are obtained in descending order of likelihood, and the speech is ranked in descending order of likelihood for the attribute “author”. When the recognition results “Ryunosuke Akutagawa”, “Dokuho Kunikida”, etc. are obtained, the speech recognition results “linear algebra and its application” with the highest likelihood are obtained for the attributes “title” and “author”, respectively. There cannot be a combination of "Ryunosuke Akutagawa" (the publication "Ryunosuke Akutagawa" written "Linear Algebra and its Applications").

Then, in the retrieval unit 3-6, the result of the voice recognition from the system control unit 3-3 and its likelihood are based on the book data relation table (FIG. 27) stored in the book database 5 described later. , A possible combination, that is, a combination of the voice recognition result “I am a cat” having the second highest likelihood in the attribute “title” and the voice recognition result “Ryunosuke Akutagawa” having the highest likelihood in the attribute “author” ( "I am a cat", "Ryunosuke Akutagawa") and are output to the system control unit 3-3.

In this case, a value obtained by adding the likelihoods of the speech recognition results "I am a cat" or "Ryunosuke Akutagawa", for example, is taken as the likelihood of this combination.

Further, in the search unit 3-6, if there are other possible combinations in the voice recognition result from the system control unit 3-3 other than the above-mentioned combinations, the system control unit 3- It is designed to output to 3.

Also, the voice recognition is performed by the recognition unit 3-4 (see FIG.
When the dictation unit 3-13 of 6) is performed, as described above, the correct voice recognition result is finally obtained. In this case, therefore, the search unit 3-6, the system control unit 3- It is only confirmed whether there is a possible combination of voice recognition results for a plurality of keywords from 3.

As described above, when the search unit 3-6 collects the voice recognition results into possible combinations, the combination results are supplied to the system control unit 3-3, and the process proceeds to step S3-91.

If there is no possible combination of the voice recognition results of each of the plurality of keywords, a message to that effect is transmitted to the terminal 2 and the process is terminated. If there is only one keyword received from the terminal 2, the process of step S3-90 is skipped and the process proceeds to step S3-91.

In step S3-91, among the combinations of the voice recognition results (including the voice recognition result when only one keyword is received from the terminal 2), the one having the highest likelihood of the combination is the terminal 2 Sent to.

As a result, in the terminal 2, the combination of the voice recognition results from the center system 1 (system control unit 3-3) is output from the display 2-7 or the voice synthesizer 2-6.

Then, step S3-92 shown in FIG.
Then, the system control unit 3-3 outputs a message asking (inquiring) whether or not the combination of the voice recognition results just transmitted is correct, which causes the message to be displayed on the display 2-7 at the terminal 2.
Alternatively, it is output from the voice synthesizer 2-6.

Furthermore, in step S3-92, it is determined whether or not the combination of the transmitted voice recognition results is correct. Here, this determination is performed as follows.

That is, if the combination of the voice recognition results output from the display 2-7 or the voice synthesizer 2-6 is correct, the user, for example, Yes / confirms the button 2-4 (FIG. 7) in the terminal 2. If the button 2-31 is operated and the combination of the voice recognition results is incorrect, the No button 2-32 of the button section 2-4 (FIG. 7) of the terminal 2 is operated, for example. As a result, an operation signal corresponding to this operation is transmitted from the terminal 2, so that the system control unit 3-3 determines whether or not the combination of voice recognition results is correct based on this operation signal. .

If it is determined in step S3-92 that the combination of voice recognition results is not correct, that is, if the user operates the No button 2-32 (FIG. 7) on the terminal 2 side, step S3- Proceeding to 93, the combination having the next highest likelihood after the combination of the voice recognition results transmitted last time is transmitted to the terminal 2, and step S3
Return to -92. As a result, in the terminal 2, the combination having the highest likelihood next to the combination of the voice recognition results being output now is output from the display 2-7 or the voice synthesizer 2-6.

If it is determined in step S3-92 that the combination of voice recognition results is not correct,
Instead of performing the process of step S3-93, for example, the changeover switch 2-36 of the button unit 2-4 (FIG. 7) in the terminal 2 is operated to activate the recognition unit 3-4 (FIG. 16) in the center system 1. Dictation section 3-13
The voice recognition may be performed by the dictation, and the user may correct the incorrect voice recognition result as described in FIGS. 22 and 23.

On the other hand, when it is determined in step S3-92 that the combination of the voice recognition results is correct, that is, on the terminal 2 side, the Yes / confirm button 2 is pressed by the user.
When -31 (Fig. 7) is operated, step S3-94
Then, the voice recognition result determined to be correct, that is, the keyword combination is output to the search unit 3-6 (FIG. 13) together with the control signal indicating that the information of the publication corresponding to the combination is searched. . As a result, the search unit 3-6 searches for information on publications corresponding to the combination of keywords supplied from the system control unit 3-3, and the search result is output to the system control unit 3-3. become.

In the system control section 3-3, step 3-
After outputting the keyword combination and the control signal to the search unit 3-6 at 94, the process proceeds to step S3-95, and it is determined whether the search result is received from the search unit 3-6.
When it is determined in step S3-95 that the search result has not been received from the search unit 3-6, step S3-
Return to 95. When it is determined in step S3-95 that the search result is received from the search unit 3-6, the process proceeds to step S3-96, and the number of publications (search number) obtained as a result of the search is the terminal 2 Sent to.

Then, at the terminal 2, the number of searches is
It is output from the display 2-7 or the voice synthesizer 2-6 so that the user can know how many publications have been searched. At this time, by operating the No button 2-32 of the button section 2-4 in the terminal 2, the following processing can be stopped and the search mode can be exited.

The system control unit 3-3 outputs the number of retrievals to the terminal 2, and then proceeds to step S3-97,
It is determined whether or not the number of searches is 0. If it is determined in step S3-97 that the number of searches is 0, the process returns to step S3-86 in FIG. 24, whereby the center system 1 is brought into a state in which a search can be performed using a new keyword. .

If it is determined in step S3-97 that the number of searches is not 0, that is, if there is one or more publications searched, the process proceeds to step S3-98 to add a new keyword. Then, a message (inquiry) asking whether to narrow down the searched publications (for example, "Do you want to narrow down?") Is displayed in Terminal 2.
Is output to step S3-99, and it is determined whether or not to narrow down the retrieved publications. Here, this determination is performed as follows.

That is, as described above, step S3-9
At 8, a message inquiring whether or not to narrow down is transmitted to the terminal 2, and at the terminal 2,
It is output from the display 2-7 or the voice synthesizer 2-6. Then, the user of the terminal 2 responds to whether or not to perform the narrowing down, depending on whether or not the narrowing down is performed, the voice recognition device 2-2, the keyboard 2-3, the button portion 2-4, or the jog shuttle portion 2
Operate -5.

Specifically, in the case of narrowing down, for example, a voice "Yes" is input to the voice recognition device 2-2, or a Yes / confirm button 2-of the button portion 2-4 (FIG. 7).
31 is operated. Further, when the narrowing down is not performed, for example, a voice “No” is input to the voice recognition device 2-2, or the No button 2-of the button unit 2-4 (FIG. 7).
32 is operated.

As a result, whether or not to narrow down from the terminal 2 in response to the operation of the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5. The operation signal shown is transmitted. In step S3-99, the system control unit 3-3 determines whether or not to narrow down the retrieved publications based on the operation signal.

If it is determined in step S3-99 that narrowing-down is to be performed, the process returns to step 3-86 in FIG. 24, a new keyword is added as described above, and the processing in steps S3-86 and thereafter is executed. Will be done again.

On the other hand, when it is determined in step S3-99 that the narrowing-down is not performed, step S3-10
In step 0, information regarding the outline of the contents of the retrieved publication information is transmitted to the terminal 2, and the process proceeds to step S3-101. In step S3-101, of the information on the retrieved publications, information about the text thereof is transmitted to the terminal 2 as necessary, and the processing ends.

The information transmitted to the terminal 2 as described above is recorded by the information recording / reproducing apparatus 2-12 in the terminal 2 (FIG. 5), for example, as described above.

In the center system 1, by inputting, for example, the voice "end" to the voice recognition device 2-2 on the terminal 2 side, the process is interrupted even during the search process. Has been done. further,
Even when the line is disconnected for some reason, the processing is interrupted and the state is returned to a state in which new access from another terminal can be accepted.

When a keyword is input in steps S3-86 and S3-87 of FIG. 24,
For example, keywords that are known to be input in advance can be omitted. That is, basically, for example, when searching for a desired article in a newspaper that is issued daily, the date of issue must always be entered as the keyword. In such cases,
In the terminal 2 or the center system 1, it is possible to set up information such that the date of issue is the current day, and this keyword, that is, the date of issue can be omitted.

Furthermore, in the above, the terminal 2
From Fig. 5, the keywords etc. were input by voice, but the keyboard 2-3 and the joc shuttle unit 2 were used.
It is also possible to input a keyword by operating -5 (when operating the jog shuttle unit 2-5, it will be described later).

Next, FIG. 26 shows the search unit 3 in the voice input type database search / transmission system 3 shown in FIG.
6 shows a detailed configuration of -6. The search unit 3-6 includes a search control unit 3-41, a search memory 3-42, and a book usage table 3
-43 and a personal database 3-44, and the search control unit 3-41 controlled by the system control unit 3-3, if necessary, a search memory 3-42,
Book Usage Table 3-43 or Personal Database 3-4
4 is referred to, or information of publications stored in the book database 5 is searched while writing information in them.

In the search memory 3-42, for example, information necessary for searching the book database 5, search results from the book database 5 (interim results of the search), search information for each user (for example, which user The user ID) etc. added to the interim result of the search are stored so that the interim result of the search can be identified.

In the book usage table 3-43, information that classifies publications according to the number of searches (search frequency) of publications, and keywords (category) (publishing) used in the search (publication) are listed. Information in which keywords (categories) (attributes of publications) are classified according to the number of times the attribute of a product is used is stored.

The personal database 3-44 contains personal information of the user (for example, information such as periodical subscription to magazines) and information relating to searches by the user such as the past usage status of the user. Is memorized.

Next, the retrieval unit 3-6 makes the book database 5
The operation of interleaving the information of the publications stored in will be described. The structure of the book database 5 will be described as a preparation for the interstage.

FIG. 27 schematically shows information on publications stored in the book database 5. As shown in the figure, the book database 5 roughly stores book data, attribute data, and a book data relation table.

The book data is composed of the data of the main text of the publication and the data of its outline (abstract). In the text data or abstract data,
A body ID or an abstract ID as a unique number is added to each publication. In addition,
For example, the body ID of the data of the body of a certain publication and the abstract ID of the data of the abstract of the publication do not have to be the same.

Attribute data is, for example, the title of a publication,
Author name, publisher, keyword (Keyword here is included in the keyword (broadly defined keyword) input at the time of search, for example, a narrowly defined keyword that means a specific word etc. described in the publication) , Fields (eg, literary books, mathematical books, etc.), date of publication, number of pages, and other attributes (categories) of the publication, each of which has a unique number. Attribute IDs (title ID, author ID, publisher ID, keyword ID, field ID, issue date ID,
The page number ID, etc.) is added. The attribute ID does not have to be the same for each publication (title, author name, publisher, keyword, field, etc. of the same publication).
The issue date and the number of pages do not have to be the same.)

The book data relationship table is composed of a book ID relationship table and a book data network table. The book ID relation table stores the correspondence between the body data, abstract data, or attribute data such as title or author name of each book, which is the body ID, abstract ID, or attribute ID of each book. A book ID as a unique number is added to the book.

Here, the book means a set of text data, abstract data, and attribute data for each publication. Therefore, the book ID can be said to be a number that identifies the publication.

By referring to the book ID relation table, the body ID, the abstract ID, or the attribute ID of each of the body text data, abstract data, or attribute data of the publication can be obtained from the book ID of the publication. It is possible to obtain the text data of the publication, the abstract data, or the attribute data.

Next, in the book data network table, the book ID, the body ID, the abstract ID, and the attribute I
The correspondence relationship of D is stored. That is, a network representation of the connection of the book ID, the body ID, the abstract ID, and the attribute ID is stored.

According to the book data network table, similar to the book ID relation table, for example, body data, abstract data, or attribute data of a body of a book having a certain book ID, abstract ID, or attribute ID, respectively. Other than that, for example, the book ID of a publication (book) written by an author with a certain author name ID, the publisher ID of the publisher of the publication written by the author, or the field ID It is possible to obtain a book ID of a publication (book) belonging to a field, an author name ID of an author who has a book in the field, and the like.

Next, the data formats of the book data and the attribute data will be described with reference to FIG. First, as shown in the figure, of the book data, for example, the text data is the data of the text itself described in the publication (hereinafter referred to as text text data) and the additional data of the book as other necessary data. Composed of.

The book additional data that composes the data of this text is the information on the table of contents of the publication, the control information necessary for displaying the text or outputting the text with synthetic sounds, the photographs, figures and graphs described in the text. It is composed of image information such as, and effect information such as music and animal calls, musical instrument sounds, and sound effects to make the reading and display effective when the sentence is read and displayed on the terminal 2. There is.

Here, the control information necessary for displaying a sentence is, for example, information such as how many characters are used in one line and where a line feed is made. Further, the control information necessary when outputting a sentence as a synthetic sound is information for controlling the intonation of the synthetic sound, controlling the utterance speed, and the like.

As described above, the book-added data can be included in the text data as well as in the abstract data. That is, the text data can be composed only of text text data, and the abstract data can be composed of text data (hereinafter referred to as synopsis text data) relating to the outline of the publication and additional data to the book. . In addition, part of the book additional data may be included in the text data, and the rest may be included in the abstract data.

Body text data in the body data, summary sentence data in the abstract data, table of contents information in the book additional data, and attribute data (author name, title, publisher, (in a narrow sense) keyword) , Field,
The date of issue (issue date), number of pages, ...) indicates phonetic symbol information, text data, kana characters, kana characters for braille, and voice signals, as well as encoded data and compressed data of these. Among the data formats, the data format including at least phonetic symbol information is stored in the book database 5.

Further, the control information in the book additional data is control data (for example, data consisting of control code).
The data format is stored in the book database 5. Further, image information such as photographs, drawings, and graphs in the book additional data is stored in the book database 5 in the data format of the image data or the data format in which the data is compressed or encoded. Further, the effect information such as music and animal calls, instrument sounds, and sound effects in the book additional data is in the data format of sound data (for example, waveform data), or a data format in which it is compressed or encoded. It is stored in the book database 5.

An identification code indicating its own data format is added to the above-mentioned data stored in the book database 5 in each data format.
It is made possible to identify what kind of data format it is.

Next, the operation of the search control unit 3-41 constituting the search unit 3-6 will be described with reference to the flowchart in FIG. First, in the search control unit 3-41, in step S3-111, it is determined whether or not the book ID is input from the system control unit 3-3.

If it is determined in step S3-41 that the book ID has been input, the flow advances to step S3-112 to refer to the book ID relationship table of the book data relationship tables stored in the book database 5 and enter it. The book data (text data or abstract data) of the book with the specified book ID, or the attribute data of the publication such as title and author name is searched from the book database 5 according to predetermined search items. Then, the process proceeds to step S3-113. In step S3-113, the search result in step S3-112 is the system control unit 3-
3 is transmitted, and the processing is ended.

On the other hand, when it is determined in step S3-111 that the book ID is not input from the system control section 3-3, the process proceeds to step S3-114, and the process shown in FIG.
As described in step S3-90 of 4, the system control unit 3-3 determines whether the attribute list which is the voice recognition result of the recognition unit 3-4 is input.

An example of the attribute list is shown in FIG.
Shown in. FIG. 30A shows step S3-86 of FIG.
In, for example, keyword (attribute name) (category)
The attribute list is shown as a recognition result of the voice uttered by the user of the terminal 2 when the question "Is the publisher?" Is asked (asked). In the attribute list, P, Q, R, ... Described on the left side are words obtained as a recognition result, and the numbers 0.
7, 0.2, 0.01, 0.01, ... Show the likelihood of words.

If it is determined in step S3-114 that the attribute list is not input, that is, if the information that is neither the attribute list nor the book ID is input,
The process proceeds to step S3-115, it is determined that there is an input error, and the process proceeds to step S3-116. Step S3-11
In 6, the error signal is the system control unit 3-3 (FIG. 13).
Is output to and the processing ends.

If it is determined in step S3-114 that the attribute list has been input, step S3.
Proceeding to -117, it is determined whether or not a plurality of attribute lists have been input. When it is determined in step S3-117 that a plurality of attribute lists are not input, step S3-118 is skipped and step S3-11 is performed.
Proceed to 9.

On the other hand, if it is determined in step S3-117 that a plurality of attribute lists have been input, then the flow proceeds to step S3-118, and the input plurality of attribute lists are as shown in FIG. 30 (b), for example. It is summarized in a table type attribute list.

That is, in step S3-118, the book data relation table stored in the book database 5 (FIG. 27).
24, the plurality of attribute lists, that is, the plurality of keywords as the voice recognition results are combined into a possible combination as described in step S3-90 in FIG.

In FIG. 30A, the publisher name "P", title "X", author name "A" as the voice recognition result.
The combination (1) has the highest likelihood (likelihood of the combination) among the combinations that can be the combinations of the keywords. Further, the combination of the publisher name “P”, the title “X”, and the author name “B” as the voice recognition result has a likelihood (likelihood of combination) of 2 among possible combinations of keywords. It is considered the second highest.

[0375] In step S3-118, after the processing of collecting a plurality of attribute lists is performed, the processing result,
That is, a table type attribute list as shown in FIG. 30B is transmitted to the system control unit 3-3, and step S3-1
In step 19, the book ID is searched after waiting for the instruction from the system control unit 3-3.

That is, in step S3-119, the system control unit 3-3 issues a control signal for instructing a search and a keyword for the search (for example, FIG. 30 (b)).
When the publisher “P”, the title “X”, the author name “A”, etc., which constitute the combination with the highest likelihood of combination in the table-type attribute list shown in FIG. The stored book data relationship table (FIG. 27) is referred to, and the received keyword I
The book ID associated with D (attribute ID) is searched.

At this time, the search control unit 3-41, if necessary, the search memory 3-42 and the book usage rate table 3-42.
43 or the stored contents of the personal database 3-44 are searched.

That is, first, the search control unit 3-41 (FIG. 26).
Then, the user ID transmitted from the terminal 2 is confirmed in step S3-1 of FIG. 14, and then a new search or continuous search (for example, step S of FIG. 25) is performed.
In 3-99, it is determined that the search is to be performed, and it is confirmed whether or not the search is performed again.

If it is a continuous search, the search memory 3
Confirmed user I stored in -42 (Fig. 26)
An intermediate result of the search to which D is added is read, and the search is performed only on the intermediate result.

If it is a new search, the search memory 3
The search is performed without referring to the stored contents of -42.

Further, regardless of new search or continuous search, it is stored in the personal database 3-44 (FIG. 26).
For example, information set in advance for each user (for example, searching only newspapers or searching only mathematical books) (hereinafter referred to as personal information) is read in correspondence with the user ID. The search is performed by limiting the search target based on the personal information.

Furthermore, the book utilization table 3-43 (FIG. 26)
Depending on the search frequency (search frequency) of the publications stored in, the information classified by the publications is read out, and the search is limited to the publications with the highest search frequency and sequentially searched. Is done. This makes it possible to speed up the search.

During the search, the search control unit 3-4
1 is a search memory 3-42, a book usage rate table 3-43,
Alternatively, when the information to be stored in the personal database 3-44 is generated, the information is stored in the search memory 3-4.
2, book usage table 3-43, or personal database 3
Each is stored in -44.

Further, in step S3-119, the search is performed while determining whether or not the number of book IDs retrieved from the book database 5, that is, the number of books (books) is m or more. Then, when it is determined that the number of searched books is m or more, that is, when the number of searched books is very large, the search is interrupted at that point, and the process proceeds to step S3-120.

That is, when the number of books having the book ID associated with the keyword ID (attribute ID) received from the system control unit 3-3 is very large, the search takes a very long time, If the system is out of order, it may mislead the user, so the search process is interrupted midway.

In step S3-119, when the search for the keyword ID received from the system control unit 3-3, that is, the book ID associated with the attribute ID is completed or the search is interrupted, the step S3-1
In step 20, the book ID obtained as a result of the search is transmitted to the system control unit 3-3 together with the number of books (the number of books in the book). In this case, in step S3-119, when the search process is interrupted midway, the information notifying that the search process was interrupted midway is also transmitted to the system control unit 3-3.

[0387] Then, proceeding to step S3-121, in preparation for the above-described continuous search, the search result is sent to the user I who has been sent from the terminal 2 at step S3-1 in Fig. 14.
It is associated with D, transferred to and stored in the search memory 3-42 (FIG. 26), and the process ends.

It should be noted that the above-described processing of collecting a plurality of attribute lists in step S3-118, that is, a plurality of keywords into possible combinations is not performed by the search unit 3-6 (search control unit 3-41). , Recognition unit 3-4
May be performed by referring to the book data relation table stored in the book database 5 (FIG. 27). In this case, the recognition unit 3-4 outputs the table-type attribute list shown in FIG. 30B to the system control unit 3-3 as a voice recognition result, and as a result, the system control is performed. When the search control is performed by the unit 3-3, the table-type attribute list is input to the search unit 3-6.

In the search unit 3-6, when the table type attribute list is input as described above, the combinations of the keywords from the highest likelihood (likelihood of combination) to the nth keyword Book ID corresponding to
Can be searched at once. Therefore,
In this case, the search result output to the system control unit 3-3 corresponds not to the book ID corresponding to a certain combination of keywords, but to each of the n combinations of keywords having a high likelihood of the voice recognition result. It becomes the book ID.

Next, referring to FIG. 31, the operation when the newspaper service is provided in step S3-5 of FIG.
This will be described with reference to the flowchart in FIG. In addition, in FIG. 31, T or H attached to the upper left of each step is
It means that the process of that step is the process of the terminal 2 or the process of the center system 1, respectively.

First, in step S3-131, the terminal 2 (FIG. 1) is started up (the power switch is turned on, etc.), the flow proceeds to step S3-132, and in step S3-1 of FIG. As described above, line connection with the center system 1 is performed (input of user ID, etc.), and the process proceeds to step S3-133. In step S3-133, the user of the terminal 2 provides the newspaper service (the center system 1 executes the newspaper service program), the voice recognition device 2-2, the keyboard 2-. 3, the button portion 2-4, or the jog shuttle portion 2-5 (FIG. 5) is operated, the process proceeds to step S3-134, and the terminal 2
From the selection signal indicating that the newspaper service was selected,
The user ID is transmitted to the center system 1.

In the center system 1, step S3-
When the selection signal and the user ID transmitted from the terminal 2 are received at 134, the system control unit 3-3 (FIG. 13)
Causes a program relating to the newspaper service to be started, and the attribute data “newspaper” is retrieved by the retrieval unit 3 in order to retrieve only the user ID and the newspaper received thereby.
-6 is sent.

Then, in step S3-135,
The search unit 3-6 confirms whether the user ID transmitted from the system control unit 3-3 is set for regular subscription. That is, in the search unit 3-6 (FIG. 26), the search control unit 3-41 causes the personal database 3
Reference is made to -44.

Then, in steps S3-136,
It is determined whether or not the subscription has been set for the user ID transmitted from the system control unit 3-3. In step S3-136, the system control unit 3
-3, if it is determined that the subscription is set for the user ID transmitted from S-3, step S3-1
Proceeding to step 38, the book ID and preset attribute data (hereinafter referred to as the setting information regarding the subscription) such as the title and the date of issue when the subscription is set are searched from the personal database 3-44, and the terminal is searched. 2 is transmitted, and the process proceeds to step S3-139.

In step S3-139, the terminal 2 receives the setting information relating to the regular subscription transmitted from the center system 1 in step S3-138, and the terminal 2 receives the book (newspaper) to which the regular subscription is set. The user is asked whether he / she wants to transfer the necessary data. Then, depending on whether or not the user transfers necessary data of the book for which the subscription is set, the voice recognition device 2-2 and the keyboard 2
-3, button section 2-4, or jog shuttle section 2-5
(FIG. 5) is operated, the process proceeds to step S3-140, and the voice recognition device 2-2, the keyboard 2-3, the button unit 2-4, or the jog shuttle unit 2 is transferred so as to transfer the data.
It is determined whether -5 has been operated.

In step S3-140, the voice recognition device 2-2 and the keyboard 2- are used to transfer the data.
If it is determined that 3, the button portion 2-4, or the jog shuttle portion 2-5 is not operated, step S3-1
Proceed to 43.

Further, in step S3-140, the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2 is transferred so as to transfer the data.
If it is determined that the operation of -5 has been performed, step S3
-141, and from the terminal 2, the book ID and the data item of the book desired to be transferred are the center system 1
Sent to.

When the center system 1 receives the book ID and the data item of the book desired to be transferred, which is transmitted from the terminal 2, step S3-142.
In the search section 3-6, in the book database 5 corresponding to the book ID and the data item received from the terminal 2.
Then, the data to be transferred is searched for, transmitted to the terminal 2, and the processing ends.

On the other hand, if it is determined in step S3-136 that the user ID transmitted from the system control unit 3-3 is not set for the regular subscription, the process proceeds to step S3-137 to perform a search. Information notifying that the subscription is not set is transmitted from the section 3-6 to the system control section 3-3, and step S3-14 is performed.
Go to 3.

In step S3-143, the terminal 2
The operation mode is set to the search mode, so that the terminal 2 operates in the search mode (details will be described later). Then, proceeding to step S3-144, the center system 1 performs a search process and a data transfer process (details will be described later) corresponding to the operation of the search mode of the terminal 2 at step S3-143.
The process ends.

Next, the operation of the terminal 2 in step S3-143 and the operation of the center system 1 in step S3-144 will be described in detail with reference to the flowchart in FIG. Note that, also in FIG. 32, as in the case of FIG. 31, T or H indicating that the processing of each step is the processing of the terminal 2 or the processing of the center system 1 is added to the upper left of each step. is there.

First, in the terminal 2, step S3-15.
1, a voice recognition device 2-2, a keyboard 2-3,
By operating the button section 2-4 or the jog shuttle section 2-5, the user waits until the user selects a category of attribute data (for example, “title” or “publisher name”). Even if a predetermined time (t seconds) has elapsed,
If no category is selected, a message prompting the user to select a category is output from the display 2-7 or the voice synthesizer 2-6.

In terminal 2, if the user does not select a category even if a message prompting the user to select a category is output from display 2-7 or voice synthesizer 2-6 multiple times. A question message asking whether or not to end the process is output from the display 2-7 or the voice synthesizer 2-6.

After that, the terminal 2 displays the question message a plurality of times on the display 2-7 or the voice synthesizer 2-.
Even if the data is output from 6, the voice recognition device 2-2, the keyboard 2-3, the button unit 2-4, or the jog shuttle unit 2 may be set so that there is no response from the user or the processing is terminated.
When -5 is operated, the line with the center system 1 is disconnected, and the process ends.

On the other hand, when the category is selected in step S3-151, the process proceeds to step S3-152, and the terminal 2 selects the attribute data (for example, the category "issue date") for the selected category. In this case, the date "○ month x day" is set by the user operating the voice recognition device 2-2, the keyboard 2-3, the button unit 2-4, or the jog shuttle unit 2-5. Wait for input. Then, similar to the case in step S3-151 described above, when the attribute data is not input, the processing ends.

If attribute data is input in step S3-152, the flow advances to step S3-153,
The user ID and the inputted attribute data are transmitted from the terminal 2 to the center system 1 together with a search instruction signal (data transfer instruction) for instructing a search for a book corresponding to the attribute data.

In the center system 1, the user ID and the attribute data from the terminal 2 are received, and step S
In 3-154, the above-mentioned attribute list (including the table-type attribute list) (FIG. 30) is created from the attribute data, and the flow proceeds to step S3-155. In the terminal 2, when the attribute data is input by operating the keyboard 2-3, the button unit 2-4, or the jog shuttle unit 2-5 other than the voice recognition device 2-2, the attribute list is displayed. The likelihood at is, for example, 1 as the highest value.
It is said that

[0408] In step S3-155, the search unit 3-6 of the center system 1 (Fig. 13) executes step S3.
The search is performed as described above based on the attribute list created in 3-154. If the attribute data transmitted from the terminal 2 is the book ID, the process of step S3-154 is skipped and step S3-154 is skipped.
In 3-155, the data corresponding to the book ID is retrieved from the book database 5.

[0409] When the search by the search unit 3-6 is completed, the flow advances to step S3-156, and the information related to the search results of the search unit 3-6 (for example, the number of searches and the book ID of the searched book) is It is transmitted to the terminal 2 via the system control unit 3-3 and the communication control unit 3-1.

At Terminal 2, step S3-157.
In, the number of search results transmitted from the center system 1 is output from the display 2-7 or the voice synthesizer 2-6, and a question message "whether to narrow down the search results" is displayed. Do you want to do? ", Display 2-7 or voice synthesizer 2
It is output from -6.

Then, in step S3-158,
The voice recognition device 2 is selected by the user so as to narrow down.
-2, the keyboard 2-3, the button portion 2-4, or the jog shuttle portion 2-5 is determined.
When it is determined in step SS3-158 that the voice recognition device 2-2, the keyboard 2-3, the button portion 2-4, or the jog shuttle portion 2-5 is operated so as to perform the narrowing down, step S3-151. Then, the process from step S3-151 is repeated again.

In step S3-158, it is determined that the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 is operated so as not to perform the narrowing down. In this case, step S3-
Proceed to step 159 to select the data to be transferred to terminal 2, display 2-7 or voice synthesizer 2
A message is output from -6, and step S3-16
Go to 0.

In step S3-160, the user operates the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 so that the data to be transferred is obtained. It is determined whether or not it is selected. If it is determined in step S3-160 that no data has been selected, step S3-1.
61 and S3-162 are skipped, and the process ends.

If it is determined in step S3-160 that data has been selected, step S3-1.
In step 61, the data transfer command is transmitted to the center system 1 together with the book ID corresponding to the data so as to transfer the data selected in step S3-160, and the process advances to step S3-162.

In step S3-162, the center system 1 receives the data transfer command and the book ID from the terminal 2, and the data associated with the book ID is read from the book database 5 and sent to the terminal 2. The data is transferred and the process ends.

By the way, in the terminal 2, by operating the jog shuttle unit 2-5 (FIG. 7), various pre-registered attribute data (herein referred to as an index) is displayed 2-7 or voice synthesized. It is output from the device 2-6 so that the desired index can be easily found when performing the above-mentioned search.

Hereinafter, with reference to the flowchart of FIG. 33, by operating the jog shuttle unit 2-5 (FIG. 7), various indexes registered in advance are displayed.
The operation when output from the display 2-7 or the voice synthesizer 2-6 will be described.

The shuttle ring 2-51 of the jog shuttle unit 2-5 (FIG. 7) can be rotated rightward or leftward, for example, up to the scale 6 at maximum. .

First, in step S2-51, the shuttle ring 2-51 is rotated to the right, the process proceeds to step S2-52, and the shuttle ring 2-51 rotated to the right in step S2-51 is rotated to the predetermined position. It is determined whether the maximum scale (maximum scale in the clockwise rotation direction) 6 is fixed for the time t seconds. If it is determined in step S2-52 that the shuttle ring 2-51 has not been fixed on the maximum scale 6 for the predetermined time t seconds, step S2-52.
Go to 2-66.

If it is determined in step S2-52 that the shuttle ring 2-51 has been fixed at the maximum scale 6 for the predetermined time t seconds, step S2-5.
3 and it is determined whether the shuttle ring 2-51 is still fixed on the maximum scale 6. Step S2
If it is determined in -53 that the shuttle ring 2-51 is fixed at the maximum scale 6, step S2
Proceeding to -54, only the first character of the index registered in advance is output from the display 2-7 or the speech synthesizer 2-6 in the lexicographical order (Aiueo order) at a predetermined number N. To be done.

That is, for example, when N is 5, the first character of the index is "a", five "i", six "u", and one "e". Four things,
If each is registered in advance, "Ah,
It is output from the display 2-7 or the voice synthesizer 2-6.

On the other hand, if it is determined in step S2-53 that the shuttle ring 2-51 is not fixed by the scale 6, the process proceeds to step S2-55, and the shuttle ring 2-51 is fixed by the scale 5. Is determined. If it is determined in step S2-55 that the shuttle ring 2-51 is fixed on the scale 5, the process proceeds to step S2-56, and only the first character of the index registered in advance is lexicographically. In order, every predetermined number n (where n <N), the display 2-7 or the voice synthesizer 2-6 outputs the data.

If it is determined in step S2-55 that the shuttle ring 2-51 is not fixed on the scale 5, the process proceeds to step S2-57, and the shuttle ring 2-51 is fixed on the scale 4. Is determined. When it is determined in step S2-57 that the shuttle ring 2-51 is fixed on the scale 4, the process proceeds to step S2-58, and the index registered in advance, for example, from the beginning to the second character, In the lexicographic order, every predetermined number M, the display 2-7 or the voice synthesizer 2-6 outputs.

On the other hand, if it is determined in step S2-57 that the shuttle ring 2-51 is not fixed to the scale 4, the flow proceeds to step S2-59, where the shuttle ring 2-51 is fixed to the scale 3. Is determined. If it is determined in step S2-59 that the shuttle ring 2-51 is fixed on the scale 3, the process proceeds to step S2-60, and, for example, from the beginning to the second character of the index registered in advance, In the lexicographical order, every predetermined number m (where m <M), the data is output from the display 2-7 or the voice synthesizer 2-6.

If it is determined in step S2-59 that the shuttle ring 2-51 is not fixed on the scale 3, the process proceeds to step S2-61, and the shuttle ring 2-51 is fixed on the scale 2. Is determined. When it is determined in step S2-61 that the shuttle ring 2-51 is fixed on the scale 2, the process proceeds to step S2-62, and the index registered in advance, for example, from the beginning to the third character, Every predetermined number L of words are output in the lexicographic order from the display 2-7 or the voice synthesizer 2-6.

On the other hand, if it is determined in step S2-61 that the shuttle ring 2-51 is not fixed to the scale 2, the flow proceeds to step S2-63, and the shuttle ring 2-51 is fixed to the scale 1. Is determined. When it is determined in step S2-63 that the shuttle ring 2-51 is fixed on the scale 1, the process proceeds to step S2-64, and the index registered in advance, for example, from the beginning to the third character, Every predetermined number l (where l <L) in the lexicographic order, the data is output from the display 2-7 or the voice synthesizer 2-6.

If it is determined in step S2-63 that the shuttle ring 2-51 is not fixed at the scale 1, the flow proceeds to step S2-65, and the shuttle ring 2-51 is fixed at the scale 0. Is determined. When it is determined in step S2-65 that the shuttle ring 2-51 is not fixed at the scale 0, the process returns to step S2-53 and step S2-
The processing from 53 is repeated.

On the other hand, if it is determined in step S2-65 that the shuttle ring 2-51 is fixed at the scale 0, that is, from the shuttle ring 2-51 rotated right in step S2-51, for example, When the shuttle ring 2-51 returns to the position of the scale 0 due to the release of the hand or the like, the process proceeds to step S2-66, and the indexes registered in advance are displayed in the dictionary 2-7 or the display 2-7. It is output from the voice synthesizer 2-6.

[0429] Then, in step S2-67, when the shuttle ring 2-51 is rotated left, the process proceeds to step S2-68, and the output of the index from the display 2-7 or the voice synthesizer 2-6 is stopped. Then, the processing ends.

Then, for example, when the enter button 2-38 of the button section 2-4 (FIG. 7) is operated, the index currently output from the display 2-7 or the voice synthesizer 2-6 is selected. It is confirmed as an index.

It should be noted that, in step S2-51, the shuttle ring 2-51 is rotated counterclockwise so that step S2-51
2-54, S2-56, S2-58, S2-60, S2
The outputs of the indexes at -62, S2-64, and S2-66 are performed in the reverse order of the lexicographical order. Further, in this case, the shuttle ring 2-51 is rotated to the right in step S2-67, so that step S
At 2-68, the output of the index is stopped.

Next, the authoring system 4 for creating and editing phonetic symbol information used for voice synthesis, which is stored in the book database 5 of FIG. 1, will be described.

Normally, when performing voice synthesis using data of a mixture of kanji and kana, the reading error of words (especially kanji), the error of accent addition, the error of position and length of pause, and the prosody addition of phrase. It is difficult to eliminate all the errors in the above, and when creating data of kanji and kana mixed sentences, OCR (optical character reader)
Even when using, etc., the reading error of the character occurs not a little. Further, when a kanji / kana mixed sentence is manually created from a publication, an input error occurs.

[0434] In addition, in addition to sentences, photographs, charts, mathematical formulas, etc. are usually described in the publications.
When the above-mentioned reading is performed by the speech synthesizer 2-6 (Fig. 5), it is necessary to convert these photographs, charts, mathematical formulas, etc. into text (text).

[0435] Further, the speech synthesizer 2-6 performs
Since the output process of the synthetic voice of reading aloud is a sequential process, in order to improve its accessibility, it is possible to read the summary of the publication and keywords only, so to speak, so to speak, and even the part desired by the user. It is necessary to be able to search by keyword or table of contents so that reading can be done immediately.

Therefore, it is necessary to create the information for performing this quick reading and searching and to include it in the data of the publication.

Therefore, in the authoring system 4, phonetic symbol information as data suitable for speech synthesis, which is composed of phonological information and prosodic information, can be created from the publication and the error can be corrected. There is. further,
The authoring system 4 is a photograph described in the publication,
Information for converting diagrams and mathematical expressions (hereinafter referred to as image information) into text (converted to text), phonetic symbol information, and for quick reading and searching (hereinafter referred to as quick reading information and search information, respectively) Has been made to be able to create.

FIG. 34 shows the detailed structure of the authoring system 4. The text input unit 4-1 converts the information about the publication, which is input in various formats, into text data of a kanji / kana mixed sentence and outputs the text data to the text editing unit 4-2. The text editing section 4-2 edits the text data of the kanji / kana mixed sentence as needed and outputs it to the text analysis section 4-3. The text analysis unit 4-3 analyzes the text data by natural language processing, and includes, for example, reading as phonological information and accents, phrases, pauses, powers, vocalization times, etc. as prosodic information. Generate phonetic symbol information.

In the phonetic symbol string editing unit 4-5, in the phonetic symbol information (phonetic symbol string data) output from the text analysis unit 4-3, word reading error, vowel devoicing, unvoiced consonant voiced. Incorrect parts (errors), accenting errors, pose position and length setting errors, phrase prosody addition errors, etc. have been corrected, and by this, phonetic symbol information that can synthesize an accurate synthesized voice. And is supplied to and stored in the book database 5.

The fast-reading / text-search-information creating unit 4-4 selects from among the information on the publications input to the text input unit 4-1.
Information other than text, that is, image information is converted into text, and this text data is output to the text analysis unit 4-3 via the text editing unit 4-2. Further, the fast-reading / text search information creation unit 4-4 creates the above-described fast-reading information and search information based on the analysis result of the text analysis unit 4-3. The fast-reading information is output to the text analysis unit 3-3 via the text editing unit 4-2, and converted into phonetic symbol information there. Further, the search information is output and stored in the book database 5 as control information.

The display 4-6 displays the information (data) output from each block constituting the authoring system 4. The keyboard 4-7 is operated when giving a command or necessary information to each block constituting the authoring system 4.

Next, FIG. 35 shows the detailed structure of the text input unit 4-1 shown in FIG. Text input section 4-1
Is an electronic text media input unit 4-11, OCR4
-12 and a communication line 4-13. The electronic text media input unit 4-11 stores data of electronic publications among publications such as books, newspapers and magazines edited and produced by publishers and newspaper publishers (hereinafter,
Electronic publication information), magnetic tape or magnetic disk,
A portable recording medium such as a magneto-optical disc or an optical disc,
Alternatively, it is input via a communication line.

The electronic text media input unit 4
-11, from the electronic publication information, the table of contents, summary of the publication,
Character data such as headings and texts are extracted and output as text data. Further, the computerized text media input unit 4-11 extracts the image information of the publication from the electronic publication information and outputs it to the display 4-6 (FIG. 34) for display.

Furthermore, the electronic text media input unit 4
-11, SGML that can describe electronic publication information when it is processed by a document processing system such as TEX, a page description language such as PostScript, and a more advanced document structure. (Stan
dard Generalized Markup Language), the description information is also extracted, and the text editing unit 4-2 is also extracted.
(FIG. 34) and the text analysis unit 4-3, and is supplied to the quick reading / text search information creating unit 4-4.

In the OCR 4-12, a publication already published in a paper medium (a publication published without being digitized) is optically read and output as text data. Text data of a publication from an information providing service company such as a personal computer communication service or a database service is input to the communication line 4-13.

Next, FIG. 36 shows the detailed structure of the text analysis unit 4-3 shown in FIG. Text analysis unit 4-3
Corresponds to a language processing unit of a so-called text-to-speech synthesis system, and extracts phonological information and prosodic information that constitute phonetic symbol information from text data of a kanji-kana mixed sentence.

The text analysis section 4-3 has a text input section 4-1 or a quick reading / reading section via the text editing section 4-2.
Data is input from the text search information creation unit 4-4. This data is the document structure extraction unit 4-3.
Input to 1.

The document structure extraction unit 4-31 extracts the table of contents, outline, and image information of the publication from the input data and outputs it to the control information extraction text analysis unit 4-37.

In this case, when the input data is described in TEX or SGML, the table of contents, outline and image information of the publication can be easily extracted. That is, data created by a document processing system such as TEX, for example, can be extracted based on this format, because charts and expressions are expressed in a predetermined format. In this case, it is possible to easily convert a chart, an expression, or the like into text data that means the contents according to a predetermined rule.

Further, the document structure extraction unit 4-31 extracts the text data of the kanji / kana mixed sentence from the input data and outputs it to the morpheme analysis unit 4-32.

The morphological analysis unit 4-32 uses the dictionary 4-33.
The word information and grammatical rules stored in advance are referred to, the text data is morphologically analyzed to be decomposed into words, and the part of speech is determined, and the control information extracting text analysis unit 4-37 and It is output to the phonetic symbol string generation text analysis unit 4-34.

That is, for example, when the text data as shown in FIG. 37 (a) is input, the morpheme analysis unit 4-32
From, the word and its part of speech as shown in FIG.
It is output as a morphological analysis result.

Phonetic symbol string generation text analysis unit 4-34
Then, first, in the reading information setting unit 4-35, addition of a word reading, vowel lengthening, vowel devoicing, unvoiced consonant voicing, and the like are performed with reference to the dictionary 4-33.

Then, the prosodic information setting unit 4-36 identifies the clauses and extracts the relation between the clauses, and further, as prosodic information, the phrase information of the sentence, the accent position, the phoneme power, the phoneme duration, and the pause. The position, length, etc. are determined with reference to the dictionary 4-33.

Here, in FIG. 37, the reading information setting section 4-3.
5 and a more detailed configuration of the prosody information setting unit 4-36. As shown in the figure, the reading information setting unit 4-35 includes a pronunciation giving unit 4-51, and the prosody information setting unit 4-5.
Reference numeral -36 is a word accent setting unit 4-61, a phrase identifying unit 4-62, an inter-phrase information extracting unit 4-63, a pose setting unit 4-64, a phrase information setting unit 4-65, and an accent information setting unit 4. -66.

The pronunciation imparting unit 4-51 refers to the dictionary 4-33 (FIG. 36), adds a reading (pronunciation) to the word from the morphological analysis unit 4-32, and further performs processing such as vowel lengthening. Is output to the word accent setting unit 4-61 of the prosody information setting unit 4-36. Word accent setting unit 4-61
Then, the information (word accent information) stored in the dictionary 4-33 (FIG. 36) is referred to, and the pronunciation imparting unit 4-
The basic accent (accent when a word is uttered by itself) of the reading of the word output from 51 is determined.

The result obtained by adding reading and basic accents to the morphological analysis result of FIG. 37 (b) is shown in FIG. 37 (c).
Shown in. In addition, in FIG. 37 (c), the readings are shown in Roman letters. Also, the symbols "'", "@", "$",
"X" and "Q" are accent positions, devoicing, consonants,
Means sound repellency and nasal drip. For example, in FIG.
In the second parenthesis in the top line of (c) (the person's ordinary noun HO'XNIX), the part of speech of the word "person" is "ordinary noun", and the reading is "HO'XNIX" (genuine). Yes, it indicates that the accent is in “HO”.

The reading and basic accent addition results (FIG. 37 (c)) are output to the accent information setting unit 4-66.

Returning to FIG. 38, the morphological analysis result (FIG. 37 (b)) is input to the phrase identifying unit 4-62 of the prosody information setting unit 4-36 via the reading information setting unit 4-35. .
In the phrase identifying unit 4-62, the morphological analysis result (see FIG.
The paragraph is identified from (b)). That is, the words are connected based on their part of speech, whereby the clause of the text data shown in FIG. 37 (a) is identified as shown in FIG. 37 (d).

The bunsetsu identification result is output to the bunsetsu inter-segment information extraction unit 4-63, where the bunsetsu is grouped based on the relation between the bunsetsus, for example, as shown in FIG. In (e), the collected clauses are shown in parentheses ()), the pose setting unit 4-64
Is output to. The pose setting unit 4-64 sets the position and length of the pose inserted between the clauses. That is, there is a possibility that a pose is inserted in FIG.
Immediately after the closing bracket among the positions outside the bracket shown in (e), the pose setting unit 4-64 determines the position of the pose based on, for example, the number of mora.

Text data in which phrases are collected (see FIG. 3)
FIG. 37 (f) shows the result obtained by setting the pose for 7 (e)). Note that in FIG. 37 (f),
Pose is represented by slash "/".

The text data for which the pose has been set (FIG. 37 (f)) is output to the phrase information setting section 4-65. In the phrase information setting unit 4-65, the structure of the sentence is obtained from the relation between the clauses of the text data, and the phrase (phrase information) in the sentence is calculated based on the structure of the sentence.
Is set. The phrase information is used for deciding a pitch pattern, for example, during voice synthesis.

The text data in which the phrase is set is
It is output to the accent information setting unit 4-66. In the accent information setting unit 4-66, the word accent setting unit 4-66.
The phonetic symbol information (phonetic symbol string) as shown in FIG. 37 (g) is generated from the output of 61 (FIG. 37 (c)) and the output of the phrase information setting unit 4-65. -5 (FIG. 34) is output to the book database 5. Here, in FIG. 37 (g), the pose is "P
AU ".

At this time, the phrase information setting section 4-6
5 also determines the position of the accent in the phrase based on the accent movement rule of the phrase and the compound word. That is, when words are connected to form a sentence, the position of the accent of the word may change depending on whether the word is uttered in the sentence or when the word is uttered independently. In the setting unit 4-66, so to speak, a correction corresponding to the change in the accent position is performed.

Returning to FIG. 36, the control information extracting text analyzing unit 4-37 of the text analyzing unit 4-3 includes a table of contents information extracting unit 4-38, an abstract information extracting unit 4-39, and reading information of figures and formulas. It is composed of a section 4-40 and a keyword information extraction section 4-41 for body text search. The table-of-contents information extraction unit 4-38, the abstract information extraction unit 4-39, or the figure / form reading information extraction unit 4-40 uses the output of the document structure extraction unit 4-31 to extract the table of contents, summary, or images of the publications. Information is extracted respectively.

When the data input to the document structure extraction unit 4-31 is described in TEX, the table of contents information extraction unit 4-38 or the figure / formula reading information extraction unit 4-38.
At 40, the structured control information of the table of contents or the structured control information of the image information is extracted based on the description of TEX, respectively.

Keyword information extraction unit for text search 4-41
Then, the appearance frequency of the word for each part of speech is aggregated from the output of the morpheme analysis unit 4-32, that is, the words forming the text data and its part of speech, and the keyword of the publication is extracted (determined) based on the aggregate value. To be done.

Specifically, as keywords, for example, nouns, proper nouns, and independent words with high frequency of occurrence are extracted.
Information indicating a keyword is added to the extracted word. Note that keywords other than these may be used as keywords.

Control Information Extraction Text Analysis Unit 4-37
Then, as described above, publication is performed by the table-of-contents information extraction unit 4-38, the abstract information extraction unit 4-39, the figure / formula reading information extraction unit 4-40, or the text search keyword information extraction unit 4-41. Table of contents, overview, image information,
Alternatively, keywords are extracted respectively, but operating the keyboard 4-7 (FIG. 34) as necessary while watching the display on the display 4-6 displaying the data of the publication, the table of contents, the outline, the image. You can also enter information or keywords.

Control Information Extraction Text Analysis Unit 4-37
The table of contents, the outline, the image information, and the keywords extracted in (4) are output to the quick reading / text search information creating unit 4-4.

FIG. 39 shows the detailed structure of the fast-reading / text search information creating section 4-4 shown in FIG. Quick reading
The text search information creation unit 4-4 includes a table of contents creation unit 4-21, an abstract creation unit 4-22, a figure / photograph / mathematical phonetic reading creation unit 4-23, and a text search keyword creation unit 4
-24.

A table of contents creating section 4-21, an abstract creating section 4-22, a figure / photograph / mathematical reading creating section 4-23,
Alternatively, in the text search keyword creation unit 4-24, the table of contents, the outline, the image information, or the keyword output from the control information extraction text analysis unit 4-37 of the text analysis unit 3-4 (FIG. 36) is predetermined. Is converted into each of the formats, is output to the book database 5 as control information, and is converted into text data that describes each content, and is converted via the text editing unit 4-2.
It is output to the text analysis unit 4-3. This gives a table of contents,
Text data that explains the content of each of the outline, image information, or keywords is converted into phonetic symbol information,
It will be stored in the book database 5.

[0473] Note that the quick reading / text search information creating unit 4-4
In the same manner as in the control information extracting text analyzing unit 4-37 of the text analyzing unit 4-3 (FIG. 36), for example, a table of contents, an outline, image information, or a keyword is extracted and created as necessary. For publications that can not be done, look at the display 4-6,
Operate the keyboard 4-7 (Fig. 34) to display the table of contents, overview,
Image information, text data corresponding to a keyword, or control information can be created.

According to the table of contents created as described above, when the table of contents is displayed or read aloud in Terminal 2 (FIG. 1), by designating a part with the table of contents, You will be able to display and read publications. In addition, the contents of the publication can be understood by displaying the summary and reading. Furthermore, according to the keyword, it is possible to know the outline of the content of the publication by displaying or reading the keyword only. Further, by designating a keyword, it becomes possible to display or read a publication from the keyword.

Next, FIG. 40 shows the detailed structure of the phonetic symbol string editing unit 4-5 shown in FIG. Phonetic symbol string editor 4
In -5, by operating the keyboard 4-7, the editing function unit 4-70 can correct the error in the phonetic symbol information (phonetic symbol string data) output from the text analysis unit 4-3. Has been done. The phonetic symbol information output from the text analysis unit 4-3 and the phonetic symbol information output from the editing function unit 4-70 in which an error has been corrected are displayed on the display 4-6. The operation of the keyboard 4-7 is performed by observing the display on the display 4-6.

That is, for example, a word reading error (for example,
Correction of morpheme extraction error, proper noun reading error, homomorphic different reading error, etc.) can be made by the display 4-
The correct reading is performed by operating the keyboard 4-7 and inputting to the editing function unit 4-70 while watching the display of 6.

When the analysis result of the text data cannot be narrowed down to one, that is, when the analysis result is ambiguous, the text analysis unit 4-3 analyzes the information indicating that the analysis result is ambiguous. It is designed to be output together with the result, and this information is displayed on the display 4-6.
Is displayed on the screen, so that the operator of the authoring system 4 (the operator of the keyboard 4-7) is alerted.

Further, the phonetic symbol information output from the text analysis unit 4-3 and the phonetic symbol information whose error has been corrected by the editing function unit 4-70 are output to the rule synthesizing unit 4-77, where they are output. , Is regularly synthesized and output as a synthesized sound. The operation of the keyboard 4-7 is performed while listening to the synthesized sound.

That is, the vowel devoicing and voiced consonant [h] voicing errors are corrected by listening to the synthetic sound output from the rule synthesizing unit 4-77 and correcting the unnatural portion.

The correct phoneme information can be easily entered from the keyboard 4-7, but it is difficult to understand the correct accent position and phrase without actually speaking. Therefore, in the phonetic symbol string editing unit 4-5, the error of the phonetic symbol information (phonetic symbol string data) in the editing function unit 4-70 can be corrected by voice.

That is, the uttered voice is the acoustic analysis unit 4-
Acoustic analysis is performed at 78, and the analysis result is output to the prosody information analysis unit 4-79. In the prosody information analysis unit 4-79,
Prosody information is generated from the acoustic analysis result, and the editing function unit 4
-70 (pose correction unit 4-74, accent correction unit 4-74
75 and the phrase correction unit 4-76).
In the editing function unit 4-70, the wrong pose, accent,
Alternatively, the phrase is corrected, for example, by replacing it with the prosody information from the prosody information analysis unit 4-79.

Specifically, by listening to the synthesized sound output from the rule synthesizing unit 4-77, an unnatural portion of the pose, accent, or phrase is found, and that portion is detected by the operator of the authoring system 4. Correctly spoken by. The correctly uttered voice is the acoustic analysis unit 4-7.
Acoustic analysis at 8. Then, the prosody information analysis unit 4-79
At, a pitch pattern is obtained from the acoustic analysis result, and a phrase component, an accent component, etc. are extracted from the pitch pattern (pitch frequency pattern). The editing function unit 4-70 corrects a pose, an accent, or an unnatural part of a phrase by the phrase component or the accent component.

The editing function unit 4-70 includes a word reading correction unit 4-71, a vowel lengthening correction unit 4-72, a phoneme devoicing / voicedness correction unit 4-73, and a pause correction unit 4-74. Accent correction unit 4-75 and phrase correction unit 4-76
Each of them corresponds to the operation of the keyboard 4-7 and the voice input to the voice analysis unit 4-78.
Corrects reading of words (kanji, alphabets, numbers, abbreviations, etc.), vowel lengthening, phonological devoicing and devoicing, pauses, accents, and phrases.

Next, referring to FIGS. 41 to 48, correction of phonetic symbol information performed by the editing function unit 4-70 will be further described. The phonetic symbol information from the text analysis unit 4-3 is displayed on the display 4-6, for example, as shown in FIG.
It is displayed as shown in 1. In the figure, the numbers in parentheses are
The kanji reading is shown, and the slash "/" shows the accent. The hyphen "-" indicates that the vowel becomes longer.

In FIG. 41, the shaded "today" means that the analysis result in the text analysis unit 4-3 was ambiguous, that is, the kanji "today" was read. It indicates that it was unknown whether "Hello" or "Today".

In this case, by operating the keyboard 4-6,
If you enter "today", the word reading correction unit 4-71
Then, as shown in FIG. 42 (a), "(ko / nichi)" is corrected to "(kyo)". Further, in this case, the vowel lengthening correction unit 4-72 or the accent correction unit 4-7.
In 5, the long sound or the accent of "(kyo)" is corrected, and thereby, "(kyo)" is corrected to "(kyo /-)" as shown in FIG. 42 (b).

Also, the text analysis unit 4-shown in FIG.
In the phonetic symbol information from No. 3, the proper noun "Haneda" is read incorrectly as "Haneda" instead of "Hata." In this case, by operating the keyboard 4-6,
The "Haneda" portion is designated, and the correct reading "Hata" is input and corrected. The portion to be corrected is displayed on the display 4- as shown in FIG.
In FIG. 6, the image is displayed with a shadow.

Furthermore, in FIG. 43, the vowel lengthening error in the vowel lengthening correction unit 4-72 indicates that the vowel lengthening error in the reading "beikoku" of "US" in the phonetic symbol information of FIG. It is shown as corrected and "become".

In FIG. 44, the vowel of “shi” ([i] in [shi]) in the phonetic symbol information of FIG. 43 is usually unvoiced. It shows that the voicing correction is performed by the phoneme devoicing / voicing correction unit 4-73. In the figure, @ is a symbol indicating unvoiced, and is described immediately before the part to be unvoiced.

Further, in FIG. 45, a pose is inserted just after "By airplane (Hiko / -ki)" in the phonetic symbol information of FIG. Part 4-74 shows the pose being inserted there. In the figure, | x | is a symbol indicating a pose and is described at the position where the pose is inserted. Also, x
Is a number that corresponds to the length of the pose.

46, in the accent correction unit 4-75, the reading "(hata)" of "Haneda" without accent (0 type accent) in the phonetic symbol information of FIG.
However, the "ha" part is accented, and "(ha /
Was corrected "(corrected to type 1 accent)
It shows the situation.

Further, FIG. 47 shows the phrase correction section 4-76.
In FIG. 46, the phonetic symbol information of FIG. 46 is shown as a set of three phrases so that a synthesized sound having a natural prosody pattern can be obtained. In the figure, [or or] is a symbol indicating the beginning or end of one phrase, and is inserted at the beginning or end of one phrase.

As described above, the phonetic symbol information shown in FIG. 41 is corrected to the phonetic symbol information having no error as shown in FIG. 48, and is output and stored in the book database 5 (FIG. 34). .

As described above, according to this information access system, in the terminal 2 (FIG. 5), the voice recognition device 2-2, the keyboard 2-3, the button section 2-4, or the jog shuttle section 2-5 is used. By operation, you can input commands and other necessary information.
Since it is output from the voice synthesizer 2-6 or the display 2-7, even a user with a physical disability can easily handle it.

Further, since the phonetic symbol information of the publication can be obtained from the center system 1, the reading service of the publication can be easily provided to the visually impaired user.

[0496]

According to the information access system of the first aspect, the database stores at least phonetic symbol information including phonological information and prosody information, and the center system stores the phonetic symbol information stored in the database. Is transmitted to the terminal, and the terminal receives the phonetic symbol information transmitted from the center system. Therefore, by performing the voice synthesis based on the phonetic symbol information, the information stored in the database can be provided to the user by the synthesized voice. Further, in this case, since the phonetic symbol information is transmitted / received, the amount of data to be transmitted / received is smaller than that in the case where voice itself is transmitted / received, and the communication cost can be reduced.

According to the information access system of claim 4, the terminal sends the keyword output from the input means to the center system. Then, the center system searches the database for phonetic symbol information regarding the publication corresponding to the keyword transmitted from the terminal, and transmits the search result to the terminal.
Therefore, by performing voice synthesis based on the phonetic symbol information as the search result, it becomes possible to provide the user with the searched information by the synthesized voice.

According to the information access system of claim 6, the keyword of the desired publication is input to the input means by voice, the center system recognizes the voice input to the input means, and the The database is searched for phonetic symbol information based on the recognition result. Therefore, desired information can be obtained only by inputting a keyword by voice without operating a keyboard or the like.

According to the information access system of claim 7, when a plurality of keywords are input to the input means by voice, the voice recognition means has a combination of voice recognition results of the plurality of keywords. It is determined whether or not to obtain, and the voice recognition result of each of the plurality of keywords is output based on the determination result. Therefore, the false recognition rate can be reduced.

According to the information access system of the eleventh aspect, a mark is added to a predetermined position of the synthetic sound output from the voice synthesizing means. Then, the voice synthesizing unit re-synthesizes the synthetic voice from the position where the mark is added and outputs the synthesized voice. Therefore, the synthesized sound of the desired portion can be immediately obtained.

According to the twelfth aspect of the information access system, the voice inputted to the voice input means is acoustically analyzed at the terminal, and the acoustic analysis result is transmitted to the center system. Then, in the center system, voice recognition is performed based on the acoustic analysis result transmitted from the terminal, and predetermined processing is performed corresponding to the voice recognition result. Therefore, as compared with the case where the audio signal itself is transmitted to the center system, the deterioration of the audio is prevented,
The amount of communication can be reduced. Further, in this case, since a large-scale voice recognition device is not required on the terminal side, the terminal can be made compact.

According to the information access system of the thirteenth aspect, the center side control means transmits the voice recognition result of the center side voice recognition means to the terminal. Therefore, it is possible to control the operation of the terminal by voice without providing a voice recognition device on the terminal side.

According to the information access system of claim 16, when the voice recognition result is erroneous, the voice is re-input to the voice input means, and the voice is based on the erroneous voice recognition result. The voice is recognized again. Therefore, the same voice is prevented from being erroneously recognized again.

According to the information access system of the seventeenth aspect, the voice corresponding to the voice recognition result of the center side voice recognition means transmitted from the center side control means to the terminal is synthesized. Therefore, even for visually impaired users,
You can check the voice recognition result.

According to the information access system of claim 19, the center side control means searches the database for information on the publication corresponding to the voice recognition result of the center side voice recognition means and sends it to the terminal. To do.
Therefore, desired information can be obtained by voice without operating a keyboard or the like.

According to the information access system of claim 20, the terminal synthesizes the voice corresponding to the information on the publication transmitted by the center side control means. Therefore, even the visually impaired user can confirm the contents of the publication.

According to the information access system of the thirty-third aspect, the voice synthesizing means corrects the synthesized voice in accordance with a predetermined auditory characteristic. Therefore, it is possible to provide a synthesized sound that is easy to hear to a deaf user.

According to the recording medium of the thirty-sixth aspect, phonetic symbol information including phonological information and prosody information is recorded. Therefore, the synthesized sound can be easily obtained.

According to the information access system of the thirty-seventh aspect, the authoring system can generate phonetic symbol information from the information about the publication and store it in the database. Therefore, it becomes possible to easily provide the information of the publication by producing a synthetic sound from the phonetic symbol information and using the voice.

According to the information access system of the thirty-ninth aspect, since the voice synthesizing means synthesizes voice from the phonetic symbol information, it is possible to confirm the phonetic symbol information created and edited by the authoring system.

According to the 47th aspect of the information access system, the prosody information of the voice input to the voice input means is extracted, and the error of the analysis result of the analysis means is corrected based on this prosody information. Therefore, the error in the analysis result can be corrected by voice without operating the keyboard or the like.

According to the information access system of claim 53, the quick-reading information creating means creates the quick-reading information for quick-reading the publication, and the database contains the quick-reading information about the publication. It is stored together with phonetic symbol information. Then, the voice synthesizing unit outputs a synthetic sound for fast-reading the publication based on the fast-reading information. Therefore, the content of the publication can be quickly grasped.

[0513] According to the information access system of claim 57, the keyword is attribute data representing attributes of publications, and the database has a correspondence relationship between attribute data and a correspondence relationship between attribute data and publications. Is stored, and the search means searches with reference to the relationship table. Therefore, the search can be performed quickly.

According to the information access system of claim 58, the publication or attribute data is provided with a unique publication identification number or attribute identification number for identifying the publication or attribute data, and the relational table shows publications. The correspondence is described by the object identification number and the attribute identification number. Therefore, the search can be performed more quickly.

According to the information access system of claim 59, the past search information storage means stores the publication identification number or the attribute identification number searched in the past, and the search means first stores the past search information storage. Search by referring to the means. Therefore, when a search that was performed in the past is searched again, the search result can be immediately obtained.

According to the information access system of claim 60, the search range is stored in the search range storage means.
The search means performs a search within the search range stored in the search range storage means. Therefore, the search can be performed quickly.

According to the information access system of claim 61, the search frequency storage means stores the frequency with which the search means searches for phonetic symbol information relating to the publication.
The search means refers to the search frequency storage means, and sequentially performs a search from groups of phonetic symbol information regarding publications with high frequency. Therefore, it is possible to immediately search for a publication that has been frequently searched in the past.

According to the information access system of claim 63, in the keyword output means,
From the head to a predetermined number of characters are output while being controlled by the output control means. Then, a desired one of the keywords output to the keyword output means is selected by the selection means. Therefore, the desired keyword can be easily input.

[0519] According to the information access system of claim 68, the center system is connected to another center system via a network, and the exchange of information with the other center system is controlled by the network control means. To be done. Therefore, it becomes possible to share information among a plurality of center systems.

[Brief description of drawings]

FIG. 1 is a diagram showing an overall configuration of an embodiment of an information access system of the present invention.

FIG. 2 is a diagram showing a detailed configuration of a network control unit 6 in FIG.

FIG. 3 is a diagram showing a network composed of a center system 1 of FIG. 1 and other center systems 1-1 to 1-7.

FIG. 4 is a CPU 6 in the network control unit 6 of FIG.
2 is a flowchart illustrating an operation of -2.

5 is a diagram showing a detailed configuration of a terminal 2 in FIG.

6 is a diagram showing a detailed configuration of a voice recognition device 2-2 of FIG.

7 is a diagram showing a detailed configuration of a button section 2-4 and a jog shuttle section 2-5 in FIG.

8 is a diagram showing a detailed configuration of a speech synthesizer 2-6 in FIG.

FIG. 9 is a flowchart illustrating an operation when reading a publication in Terminal 2 of FIG. 5;

10 is a flowchart illustrating an operation when a publication is displayed on display 2-7 in terminal 2 of FIG.

11 is a schematic diagram illustrating a printer 2 in the terminal 2 of FIG.
8 is a flowchart illustrating an operation when printout of a publication is performed from No. 8.

12 is a flowchart illustrating an operation when reproduction is performed by the audio recording / reproducing device 2-13 in the terminal 2 of FIG.

FIG. 13 is a diagram showing a detailed configuration of a voice input type database search / transmission system 3 of FIG.

FIG. 14 is a flowchart illustrating main control of a system control unit 3-3.

FIG. 15 is a flowchart illustrating control of the system control unit 3-3.

16 is a diagram showing a detailed configuration of a recognition unit 3-4 of FIG.

17 is a diagram showing a detailed configuration of a word recognition unit 3-12 in FIG.

18 is a flowchart illustrating the operation of the word recognition unit 3-12 in FIG.

19 is a diagram showing a detailed configuration of the dictation unit 3-13 of FIG.

20 is a flowchart illustrating an operation of the control unit 3-3 when dictation is performed in the dictation unit 3-13 in FIG.

21 is a flowchart illustrating the operation of the dictation unit 3-13 of FIG.

FIG. 22 is a flowchart illustrating an operation of the terminal 2 when the dictation result of the dictation unit 3-13 is corrected.

FIG. 23 is a diagram for explaining an operation when the dictation result of the dictation unit 3-13 is corrected in the terminal 2.

FIG. 24 is a system control unit 3 when a keyword is input by voice and the book database 5 is searched.
3 is a flowchart illustrating an operation of No. -3.

FIG. 25 is a flowchart following the flowchart of FIG. 24.

FIG. 26 is a diagram showing a detailed configuration of a search unit 3-6 in FIG.

27 is a diagram showing a detailed configuration of the book database 5 in FIG.

28 is a diagram for explaining the data format stored in the book database 5 of FIG. 27. FIG.

29 is a flowchart illustrating the operation of the search control unit 4-41 in FIG.

FIG. 30 is a diagram showing an attribute list.

FIG. 31 is a flowchart illustrating an operation in the center system 1 when a newspaper service is provided.

32 is a flow chart of FIG. 31, step S3-1.
43 is a flowchart for explaining the details of the processing of S43 and S3-144.

FIG. 33 is a flowchart illustrating an operation when a preregistered index is output from display 2-7 or voice synthesizer 2-6 by operating jog shuttle unit 2-5.

34 is a diagram showing a detailed configuration of the authoring system 4 in FIG.

35 is a diagram showing a detailed configuration of the text input unit 4-1 of FIG. 34.

36 is a diagram showing a detailed configuration of the text analysis unit 4-3 of FIG. 34.

37 is a diagram for explaining the operation of the text analysis unit 4-3 of FIG. 34.

FIG. 38 is a phonetic symbol string generation text analysis unit 4 of FIG. 36;
It is a figure which shows the detailed structure of -34.

39 is a rapid reading / text search information creating unit 4-4 in FIG.
It is a figure which shows the detailed structure of.

40 is a diagram showing a detailed configuration of a phonetic symbol string editing unit 4-5 in FIG. 34.

41 is a diagram for explaining the operation of the phonetic symbol string editing unit 4-5 in FIG. 34.

42 is a diagram for explaining the operation of the phonetic symbol string editing unit 4-5 in FIG. 34.

FIG. 43 is a diagram for explaining the operation of the pronunciation symbol string editing unit 4-5 in FIG. 34.

44 is a diagram for explaining the operation of the phonetic symbol string editing unit 4-5 in FIG. 34.

45 is a diagram for explaining the operation of the phonetic symbol string editing unit 4-5 in FIG. 34.

FIG. 46 is a diagram for explaining the operation of the phonetic symbol string editing unit 4-5 in FIG. 34.

47 is a diagram for explaining the operation of the pronunciation symbol string editing unit 4-5 in FIG. 34.

48 is a diagram for explaining the operation of the phonetic symbol string editing unit 4-5 in FIG. 34.

[Explanation of symbols]

 1 Center system 2 Terminal 3 Voice input type database search / transmission system 4 Authoring system 5 Book database

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G10L 3/00 561 E 9379-5H G 9379-5H (72) Inventor Kazuo Ishii Kita Shinagawa-ku, Tokyo Shinagawa 6-735 Sony Corporation (72) Inventor Yusuke Iwahashi 6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo Inside Sony Corporation (72) Inventor Yasuhiko Kato 6 Kita-Shinagawa, Shinagawa-ku, Tokyo Chome 7-35 In Sony Corporation (72) Inventor Hiroaki Ogawa 6-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Inside Sony Corporation (72) Inventor table Masanori 6-7 Kita-Shinagawa, Shinagawa-ku, Tokyo No. 35 Sony Corporation (72) Inventor Kazuo Watanabe 6-7 Kita-Shinagawa, Shinagawa-ku, Tokyo No. 35 Sony Corporation (72) Inventor Katsuki Minamino 6-7 Kita-Shinagawa, Shinagawa-ku, Tokyo No. 35 Sony over within Co., Ltd. (72) inventor Koji Asano Tokyo, Shinagawa-ku, Kita 6-chome No. 7 No. 35, Sony over the Corporation

Claims (69)

[Claims] 1. An information access system comprising a center system having a database storing information, and a terminal connected to the center system by a communication line to access information in the database, wherein the database is phonological information and prosody. At least phonetic symbol information including information is stored, the center system transmits phonetic symbol information stored in the database to the terminal, and the terminal receives the phonetic symbol information transmitted from the center system. An information access system characterized by receiving the. 2. The information access system according to claim 1, wherein the terminal includes a voice synthesizing unit that synthesizes a voice from the phonetic symbol information. 3. The information access system according to claim 2, wherein the phonetic symbol information relates to a publication. 4. The terminal further comprises input means for inputting a keyword of the desired publication, and transmitting the keyword output from the input means to the center system, wherein the center system includes the terminal. 4. A search unit that searches the database for phonetic symbol information corresponding to the publication that corresponds to the keyword transmitted from the database, and transmits the search result of the search unit to the terminal. Information access system described in. 5. The center system transmits, to the terminal, phonetic symbol information regarding the publication, which is a search result of the search means, regarding the outline thereof. Information access system. 6. A keyword of the desired publication is input by voice to the input means, and the center system further includes voice recognition means for recognizing the voice input to the input means. The information access system according to claim 4 or 5, wherein the search unit searches the database for phonetic symbol information regarding the publication based on a recognition result of the voice recognition unit. 7. The voice recognition means includes the input means,
When a plurality of the keywords are input by voice, it is determined whether or not there is a combination of the voice recognition results of the plurality of keywords, and based on the determination result, the voice recognition results of the plurality of keywords. 7. The information access system according to claim 6, wherein the information access system outputs. 8. The input means comprises a keyboard, buttons,
The information access system according to claim 4, wherein the information access system comprises a jog dial or a shuttle ring. 9. The information access system according to claim 1, wherein the terminal further comprises recording means for recording the phonetic symbol information on a recording medium. 10. The information access system according to claim 9, wherein the terminal further comprises reproducing means for reproducing the phonetic symbol information recorded in the recording medium. 11. The terminal further comprises mark adding means for adding a mark to a predetermined position of the synthesized sound output from the voice synthesizing means, and the voice synthesizing means, from the position where the mark is added, The information access system according to claim 2, wherein the synthesized voice is resynthesized and output. 12. An information access system comprising a center system and a terminal connected to the center system via a communication line, wherein the terminal is a voice input unit for inputting voice, and a voice input to the voice input unit. Acoustic analysis means for acoustic analysis of, the acoustic analysis result of the acoustic analysis means is transmitted to the center system, the center system, based on the acoustic analysis result of the acoustic analysis means transmitted from the terminal. An information access system comprising: a center side voice recognition means for performing voice recognition; and a center side control means for performing a predetermined process in response to a voice recognition result of the center side voice recognition means. 13. The information access system according to claim 12, wherein the center-side control means transmits the voice recognition result of the center-side voice recognition means to the terminal. 14. The center side voice recognition means performs voice recognition by dictation, and the terminal is erroneously recognized among the voice recognition results of the center side voice recognition means transmitted from the center side control means. 14. The information access system according to claim 13, further comprising voice recognition result correction means for correcting a part, and transmitting the correction result by the voice recognition result correction means to the center system. 15. The information access system according to claim 14, wherein the voice recognition result correction means is a jog dial or a shuttle ring. 16. The center side voice recognition means performs voice recognition by dictation, and when the center side voice recognition means's voice recognition result transmitted from the center side control means is incorrect, the center side voice recognition means The voice is re-input, and the center-side voice recognition unit recognizes the voice re-input to the voice input unit again based on the incorrect voice recognition result. Item 13. The information access system according to Item 13. 17. The terminal further comprises a voice synthesizing unit for synthesizing a voice corresponding to a voice recognition result of the center side voice recognizing unit transmitted from the center side control unit. 16. The information access system according to any one of 16. 18. The terminal performs a predetermined process corresponding to the terminal side voice recognition means for performing voice recognition based on the acoustic analysis result of the acoustic analysis means, and the voice recognition result of the terminal side voice recognition means. The information access system according to any one of claims 12 to 17, further comprising terminal-side control means for performing the above. 19. The center system further comprises a database that stores information about publications, and the center-side control means relates to the publications corresponding to a voice recognition result of the center-side voice recognition means. 13. The information access system according to claim 12, wherein information is retrieved from the database and transmitted to the terminal. 20. The information access system according to claim 19, wherein the terminal further comprises a voice synthesizing unit that synthesizes a voice corresponding to the information on the publication transmitted by the center-side control unit. . 21. The terminal transmits a range of vocabulary to be subjected to voice recognition to the center system, and the center-side voice recognition means performs voice recognition within the range of the vocabulary transmitted from the terminal. 21. The information access system according to claim 12, wherein the information access system is performed. 22. The information access system according to claim 12, wherein the terminal compresses the acoustic analysis result of the acoustic analysis means and transmits the result to the center system. 23. The information access system according to claim 19, wherein the terminal further comprises a braille display or a braille printer that outputs information about the publication transmitted by the center-side control means. 24. The information access system according to claim 12, wherein the voice is uttered in word units or syllable units. 25. The terminal according to claim 2, further comprising recording means for recording the synthesized sound synthesized by the voice synthesizing means on a recording medium.
Information access system described in. 26. The information access system according to claim 2, wherein the terminal further comprises a synthetic sound control means for controlling a synthetic sound synthesized and output by the speech synthesis means. 27. The information access system according to claim 26, wherein said synthetic voice control means controls the voice quality of said synthetic voice. 28. The information access system according to claim 26, wherein the synthetic voice control means controls a speech rate of the synthetic voice. 29. The information access system according to claim 26, wherein the synthesized voice control means controls a voice pitch of the synthesized voice. 30. The information access system according to claim 26, wherein the synthetic sound control means is a jog dial or a shuttle ring. 31. The information access system according to claim 25, wherein the terminal further comprises reproducing means for reproducing the synthesized sound recorded in the recording medium. 32. The information access system according to claim 31, wherein the reproduction means reversely reproduces the synthetic sound. 33. The voice synthesizer outputs the synthesized voice,
21. The information access system according to claim 2, further comprising a characteristic correction unit that corrects in accordance with a predetermined auditory characteristic. 34. The information access system according to claim 33, wherein the predetermined auditory characteristic is changeable. 35. The information access system according to claim 33, wherein the characteristic correction unit is a filter. 36. A recording medium on which phonetic symbol information including phonological information and prosody information is recorded. 37. The information access system according to claim 1, wherein the center system further comprises an authoring system that creates the phonetic symbol information from information about publications and stores the phonetic symbol information in the database. 38. The information access system according to claim 37, wherein the authoring system includes an editing unit that edits the phonetic symbol information. 39. The information access system according to claim 37, wherein the authoring system includes a voice synthesizing unit that synthesizes a voice from the phonetic symbol information. 40. The information according to claim 37, wherein the authoring system includes an analyzing unit that analyzes the text data of the publication and generates the phonetic symbol information based on the analysis result. Access system. 41. The authoring system corrects the reading error of the word when the reading result of the word in the text data of the publication is incorrect in the analysis result of the analyzing unit. The information access system according to claim 40, further comprising: 42. The authoring system, when the analysis result of the analysis means includes a vowel devoicing error in the text data of the publication, devoicing for correcting the vowel devoicing error. The information access system according to claim 40 or 41, further comprising error correction means. 43. The accent error correction means for correcting the accent error of the word when the authoring system includes an error in the accent of the word in the text data of the publication in the analysis result of the analyzing means. 43. The method further comprising:
Information access system described in any of. 44. The authoring system corrects the error in the position of the pose, if the result of analysis by the analysis unit includes an error in the position of the pose inserted in the text data of the publication. 44. The information access system according to claim 40, further comprising position error correction means. 45. The authoring system further comprises phrase error correction means for correcting an error in the phrase when the analysis result of the analysis means includes an error in the phrase in the text data of the publication. The information access system according to any one of claims 40 to 44, wherein: 46. The authoring system corrects an error in lengthening a vowel when the analysis result of the analyzing unit includes an error in lengthening a vowel in the text data of the publication. 46. The information access system according to claim 40, further comprising error correction means. 47. The authoring system comprises: voice input means for inputting voice; extraction means for extracting prosody information of the voice input to the voice input means; prosody information of the voice extracted by the extraction means. The information access system according to claim 40, further comprising: a correction unit that corrects an error in the analysis result of the analysis unit based on the above. 48. The information access system according to claim 40, wherein the authoring system further comprises character reading means for reading characters from the publication and generating the text data. 49. The authoring system includes a voice synthesizing unit for synthesizing a voice from an output of the character reading unit, and an error in reading a character from the publication by the character reading unit, output from the voice synthesizing unit. 49. The information access system according to claim 48, further comprising a correction unit that corrects based on a synthesized sound. 50. The authoring system converts graphic data of the publication into text data that describes the publication, and the analysis unit further analyzes the text data that describes the graphic data. The information access system according to claim 40. 51. The information access system according to claim 40, wherein the text data of the publication includes information related to an outline of description items of the publication. 52. The authoring system according to claim 40, further comprising a summary input means for inputting a summary of description items of the publication as text data of the publication to the analysis means. Information access system. 53. The center system further comprises a quick-reading information creating unit for creating quick-reading information for quick-reading the publication, wherein the database includes the pronunciation of the publication. 4. The information access system according to claim 3, wherein the voice synthesizing means outputs a synthetic sound for fast-reading the publication based on the fast-reading information, which is stored together with the symbol information. 54. The fast-reading information comprises a keyword extracted from the publication.
Information access system described in. 55. The information access system according to claim 53, wherein the fast-reading information is table-of-contents information of the publication. 56. The information access system according to claim 55, wherein the center system further comprises a table of contents creating means for creating table of contents information of the publication. 57. The keyword is attribute data representing an attribute of the publication, and the database includes a relationship table representing a correspondence relationship between the attribute data and a correspondence relationship between the attribute data and the publication. The information access system according to claim 4, wherein the information is stored, and the search unit searches by referring to the relation table. 58. The publication or attribute data is provided with a unique publication identification number or attribute identification number, respectively, for identifying the publication or attribute data, and the relational table includes the publication identification number and the attribute identification number. 58. The information access system according to claim 57, wherein the correspondence is described. 59. The center system further comprises past search information storage means for storing the publication identification number or the attribute identification number retrieved in the past, and the search means initially includes the past search information storage means. 59. The information access system according to claim 58, wherein the search is performed by referring to the information access system. 60. The center system further comprises search range storage means for storing a search range, wherein the search means performs a search within the search range stored in the search range storage means. The information access system according to claim 4. 61. The center system includes a search frequency storage unit that stores a frequency with which the phonetic symbol information regarding the publication is searched by the search unit, and the search unit refers to the search frequency storage unit, The information access system according to claim 4, wherein a search is sequentially performed from a group of phonetic symbol information relating to the publication having the high frequency. 62. The attribute data is voice-recognized, and the search means sequentially searches phonetic symbol information relating to the publication corresponding to the attribute data having a high likelihood of the result of the voice recognition. 58. The information access system according to claim 57, which is performed. 63. The input means outputs a keyword from the beginning up to a predetermined number of characters among the keywords, an output control means for controlling output of the keyword from the keyword output means, and the keyword output The information access system according to claim 4, further comprising a selection unit that selects a desired one of the keywords output from the unit. 64. The information access system according to claim 57, wherein the attribute data is a title of the publication, a publisher name, an author name, a publication date, the number of pages, or a field. 65. The terminal according to claim 1, further comprising broadcast receiving means for receiving an emergency broadcast regarding emergency alert, and notifying means for notifying the emergency alert. Information access system described in Crab. 66. The center system and the terminal perform wireless communication with each other.
5. The information access system according to any one of 5 and 37 to 65. 67. The terminal according to claim 1, wherein the terminal operates with a battery as a power source.
The information access system according to any one of 7 to 66. 68. The center system is connected to another center system via a network, and further comprises network control means for controlling exchange of information with the other center system. The information access system according to any one of 1 to 35 or 37 to 67. 69. The information access system according to any one of claims 1 to 35 or 37 to 68, wherein the center system is connected to a plurality of the terminals.