JPH07210185A - Reading information preparing device and reading device - Google Patents

Abstract

PURPOSE:To make it possible to immediately listen only a necessary part out of the reading voice of a text by including information for obtaining the phonemic information of the text and reading control information for the fast reading of the text in reading information. CONSTITUTION:A text information storing part 2 stores text information such as a morpheme analytical result outputted from a language processing part 1 and reading control information and stores also reading control information supplied from a reading information applying part 7 and inputted from an input device 6 by a user. A reading control part 3 controls the reading of the text information from the storing part 2 based upon the reading control information stored in the storing part 2 or a control signal inputted from the device 6. Thereby text information necessary for reading is supplied to a phonemic/ rhythmic information generating part 4. The generating part 4 generates phonemic information and rythmic information and outputs these information to an acoustic processing part 5 and only necessary voice element data are outputted from a speaker as a continuous voice waveform.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is suitable for creating reading information suitable for performing speech synthesis from a text sentence containing kanji and kana in a publication such as a newspaper, a book, a magazine, that is, reading a publication. The present invention relates to a device and a reading device.

[0002]

2. Description of the Related Art Conventionally, for example, newspapers as type information,
As a reading device for reading a text sentence that digitizes a mixture of kanji and kana in publications such as books and magazines, for example, "Sentence Speech Synthesis Based on Fundamental Frequency Pattern Generation Process Model", Hirose, Fujisaki, Kawai, Yamaguchi , IEICE Transactions, Vol. J72-A No. 1 1989 and the like are known.

This reading device is, for example, as shown in FIG. 7, a language processing section 21, a prosody processing section 22, and a sound processing section 2.
First, the language processing unit 21 language-processes the input text sentence, and creates (extracts) information about its reading, that is, phoneme information. Then, the prosody information processing unit 22 extracts prosody information (information about the prosody of the text sentence), and the acoustic processing unit 23 performs speech synthesis based on only the phonological information and the prosody information. Then, the synthesized sound generated in this way is output from a speaker (not shown), so that the text sentence is read aloud.

[0004]

By the way, in the reading device, since the input text sentence is processed sequentially to read aloud, when the text sentence is composed of many characters, It took time to read aloud, so it was difficult to grasp the contents of the entire text sentence in a short time.

Therefore, there is a method of increasing the reading speed. However, if the reading speed is too high, the reading voice (synthetic voice) becomes difficult to hear. Therefore, it is difficult to greatly reduce the reading time by this method. It was

Further, conventionally, it was not possible to search for a desired portion to be read and to read only that portion. Therefore, the user has to listen to the read voice of the unnecessary portion, and therefore the user has to listen to it. There was a task that made me feel annoyed.

The present invention has been made in view of such a situation, and makes it possible to instantly listen to only the necessary portion of the reading voice of the text sentence, and further, the content of the text sentence in a short time. Is to be able to grasp.

[0008]

A reading information creating apparatus according to claim 1 is a reading information creating apparatus for creating reading information for reading a text sentence, wherein the reading information is at least a text sentence. It is characterized in that it includes information for obtaining phonological information and reading control information for quickly reading a text sentence.

According to a second aspect of the invention, the recitation information creating device is characterized in that the recitation control information includes random access information indicating a predetermined position of the text sentence.

According to the third aspect of the present invention, the reading information creating device is characterized in that the reading control information includes table of contents information indicating a table of contents of the text sentence.

According to a fourth aspect of the present invention, there is provided the reading information creating apparatus which creates the table-of-contents information based on the boundaries of chapters, sections, or paragraphs of a text sentence.

According to another aspect of the present invention, the reading information creating apparatus detects a paragraph of a text sentence based on a line feed code in the text sentence, and detects a chapter or section boundary of the text sentence by the number of characters included in the paragraph. It is characterized in that it is detected based on a number or a word representing the beginning of a chapter or section, and table-of-contents information is created based on the boundary of this chapter or section.

According to a sixth aspect of the present invention, there is provided a reading information creating apparatus which, when the text structure information is added to the text sentence, creates the table-of-contents information based on the structure information.

An aloud reading information creating apparatus according to a seventh aspect is characterized by comprising a table of contents information input means (for example, a table of contents button 14 shown in FIG. 2) for inputting table of contents information.

According to the eighth aspect of the invention, the recitation information creating device is characterized in that the recitation control information includes a specific word in the text sentence.

An aloud reading information creating apparatus according to a ninth aspect comprises word input means (for example, the keyboard 17 shown in FIG. 2) for inputting a word, and the word input to the word input means is
It is characterized in that a specific word is extracted from a text sentence.

The reading information creating apparatus according to claim 10 is
It is characterized in that a text sentence is subjected to morphological analysis, and a word having a predetermined morpheme is extracted as a specific word from the text sentence.

The reading information creating apparatus according to claim 11 is
Morphological analysis of text sentences, and based on the analysis results,
The feature is that the frequency of occurrence of an independent word in a text sentence is obtained, and the high frequency is extracted as a specific word from the text sentence.

The reading information creating apparatus according to claim 12 is
Morphological analysis of text sentences, and based on the analysis results,
The appearance frequency of an independent word in a text sentence is obtained, and the low frequency is extracted from the text sentence as a specific word.

According to a thirteenth aspect of the present invention, in the reading device for reading the text sentence based on the reading information of the text sentence, the reading information includes at least information for obtaining phonological information of the text sentence and the text sentence. And read aloud control information for quick reading.

According to a fourteenth aspect of the present invention, in the reading device, the reading control information includes random access information indicating a predetermined position of the text sentence, and the predetermined position of the text sentence is accessed based on the random access information. Characterize.

A reading device according to a fifteenth aspect is a table information input means for inputting table of contents information representing a table of contents of a text sentence as reading control information (for example, a table of contents button 1 shown in FIG. 2).
4) is provided.

The reading device according to a sixteenth aspect is characterized in that the reading control information includes table of contents information indicating a table of contents of the text sentence.

A reading device according to a seventeenth aspect is characterized in that only the table of contents information is read.

The reading device according to the eighteenth aspect is characterized in that the table-of-contents information includes boundary information indicating boundaries of chapters, sections, or paragraphs of the text sentence.

A reading device according to a nineteenth aspect is characterized in that only the head portion of a chapter, section or paragraph of a text sentence is read based on boundary information.

A reading device according to a twentieth aspect is characterized in that the chapters, sections, or paragraphs of the text sentence are read in a predetermined order based on the boundary information.

According to a twenty-first aspect of the present invention, there is provided a reading device according to the boundary information during reading of a text sentence, based on boundary information.
It is characterized by informing the boundaries of sections or paragraphs.

The reading device according to a twenty-second aspect is characterized in that the reading control information includes a specific word in the text sentence and reads only the specific word or the specific word and a portion before or after the specific word. And

A reading device according to a twenty-third aspect is provided with a random access information input means (for example, an access point button 13 shown in FIG. 2) for inputting random access information indicating a predetermined position of a text sentence as reading control information. It is characterized in that a predetermined position of the text sentence is accessed based on the random access information.

In the reading device according to the twenty-fourth aspect, the reading control information is random access information indicating a predetermined position of the text sentence, table of contents information indicating the table of contents of the text sentence, or a specific word in the text sentence. The feature is that the text sentence is accessed and read aloud based on the control information.

A reading device according to a twenty-fifth aspect is characterized in that the reading of the text sentence is performed in the direction from the beginning to the end or from the end to the beginning.

[0033]

In the reading information creating apparatus of the present invention, reading information is created from a text sentence. Then, the reading information includes at least information for obtaining phonological information of the text sentence and reading control information for quick reading of the text sentence. Therefore, when reading a text sentence by using the reading information, the content of the text sentence can be grasped in a short time by performing the quick reading of the text sentence based on the reading control information. .

In the recitation device of the present invention, the recitation information of the text sentence is included based on the recitation information of the text sentence including at least the information for obtaining the phoneme information of the text sentence and the recitation control information for quickly reading the text sentence. Read aloud. Therefore, the text sentence can be quickly read based on the reading control information, and thus the content of the text sentence can be grasped in a short time.

[0035]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing the configuration of an embodiment of a reading device to which the present invention is applied. The language processing unit 1 is a text sentence (hereinafter, referred to as a text sentence) containing a mixture of digitized kanji and kana in a publication such as a newspaper, a book, or a magazine.
When is input, the text sentence is subjected to, for example, morphological analysis, etc. to create (extract) information (text information) necessary for normal reading, and at the same time, a predetermined specific word is extracted from the text sentence. It is extracted and used as reading control information (quick reading information for performing quick reading of text sentences). In addition, table of contents information representing the table of contents is created from the text sentence, and this is used as reading control information.

The extraction method for extracting a specific word (specific word) from a text sentence is set based on a specific word extraction method selection signal supplied from a user information input device (hereinafter referred to as an input device) 6. It is designed to be.
Also, if there is a table of contents in the text, the table of contents information is created using it.

The text information storage unit with reading control information (hereinafter referred to as storage unit) 2 is composed of, for example, a semiconductor memory, and stores text information such as a morphological analysis result output from the language processing unit 1 and reading control information. Remember. Further, the storage unit 2 is also configured to store the recitation control information supplied from the recitation information adding unit 7 and input by the user from the input device 6.

The reading control unit 3 controls reading of text information from the storage unit 2 based on the reading control information stored in the storage unit 2 or the control signal from the input device 6. As a result, the phoneme / prosodic information generation unit (hereinafter, referred to as a generation unit) 4 is supplied with text information necessary for reading in a predetermined order.

The generation unit 4 uses, for example, a method similar to the conventional one,
That is, for example, the so-called Fujisaki model or point-pitch model is used to generate phonological information (reading) and prosody information corresponding to the text information from the reading control unit 3 and output it to the acoustic processing unit 5. There is.

The sound processing section 5 is composed of, for example, a synthesizer for performing speech synthesis by a method similar to the conventional one, that is, a speech synthesizer by waveform segment connection, a so-called formant synthesizer, or the like. When the voice processing unit 5 is, for example, a voice synthesizer based on waveform segment connection, the voice processing unit 5 uses, for example, CV or CV /
The speech unit data (eg, digitized waveform data) necessary for the regular speech synthesis is stored in units such as VC. First, it is necessary based on the phoneme information and the prosody information from the generation unit 4. Connect the voice unit data so as to form a continuous voice waveform. Then, the voice waveform is subjected to necessary processing such as D / A conversion processing, and output as a synthesized sound from a speaker (not shown).

The voice processing unit 5 may, if necessary,
The speech waveform described above is subjected to filtering processing by, for example, a low-pass or high-frequency enhancing digital filter having a predetermined number of taps to enhance the spectrum, thereby making it possible to make the synthesized speech easy to hear. .

Further, in the voice processing unit 5, the formant of the voice unit data can be emphasized as necessary so that the synthesized voice is easy to hear.

The input device 6 is, for example, as shown in FIG.
A jog shuttle 11 including a shuttle ring 11a and a jog dial 11b that are integrally formed, and a selection button 1
2, an access point button 13, a table of contents button 14, an interruption button 15, a reading button 16, a keyboard 17, a liquid crystal display 18, and the like.

The jog shuttle 11, the selection button 12, the access point button 13, the table of contents button 14, the interruption button 15, the reading button 16, or the keyboard 17 is used to control the operation of the apparatus or to add desired reading control information. It is operated by the user when he or she specifies, the part (position) to be read aloud, the order, and other necessary information.

Further, the liquid crystal display 18 is adapted to display, for example, a message for prompting the user to input, a selection branch for allowing the user to select a predetermined item, and other necessary information.

Next, the operation will be described. First, when a text sentence is input to the language processing unit 1, there,
For example, the processing is performed according to the flowchart shown in FIG. That is, in the language processing unit 1, first, in step S1, it is determined whether or not the input text sentence includes the structure information (whether or not the structure information is described in the input data).

Here, when the text sentence is processed by a document processing system such as TEX, or in a page description language such as PostScript,
SGML that can describe the structure of more advanced documents
In the case of being expressed by such as (Standard Generalized Markup Language), the text sentence is described in a structured manner.

That is, when the text sentence is, for example, in TEX, the chapter or section is "\ chapte".
It is designed to start with structural information such as "r" or "\ section".

Therefore, in step S1, a search for "\ chapter", "\ section", or the like is performed from the text sentence, and if it can be searched, it is determined that the text sentence includes structural information. If the text cannot be searched, it is determined that the text sentence does not include structural information.

When it is determined in step S1 that the text sentence includes structural information, the process proceeds to step S2, and based on the structural information, the boundaries of the chapters or sections of the text sentence (for example, the chapters or sections of the text respectively). Position) is detected. That is, in step S2, if the text sentence is, for example, in TEX, “¥ ch
description position ("\ chapter" or "\ section", respectively)
The position of the word immediately following each n) will be detected.

On the other hand, when it is determined in step S1 that the text sentence does not include structural information, step S3
Then, the boundary of each chapter or section of the text sentence is detected as follows. That is, each of the paragraphs at the beginning of the chapter or section is, for example, “Chapter 1” or “Chapter II
(○○○ is a title representing the content of a chapter) or "Section 1" or "Section II XXX" (XXXXX is a title representing the content of a section) Further, it is common to include a numeral representing a chapter or section number and a character such as "chapter" or "section".

Therefore, in step S3, first, the boundaries of the paragraphs of the text sentence (in this case, the positions of the beginnings of the respective paragraphs of the text) are detected. This is done, for example, by searching for a line feed code in the text sentence. After the paragraph is detected, the number of characters included in that paragraph (the number of characters between the line feed code at the end of the paragraph immediately preceding the paragraph and the line feed code at the end of the paragraph) is counted, and that is the specified value. It is determined whether or not the following. If it is determined that the number of characters included in a paragraph is less than or equal to a specified value, the paragraph contains a number and, for example, the characters “chapter” or “section”
It is determined whether and are included.

Then, when it is determined that the paragraph includes the numeral and the character "chapter", the paragraph is regarded as the first paragraph of the chapter of the text sentence, and the position of the beginning is the chapter. Is detected as the boundary of. Also, when it is determined that the paragraph includes the number and the character “section”, the paragraph is regarded as the beginning paragraph of the section of the text sentence, and the position of the beginning is the section boundary. To be detected.

After the boundary of each chapter or section of the text sentence is detected in step S2 or step S3, the process proceeds to step S4 and is used as the table of contents information.

In step S4, if there is a title (text character string) representing the content of each chapter or section in the portion after the boundary (head) of each chapter or section of the text sentence, that is also set. It can be set as table of contents information.

Further, the table of contents information may include the boundary (beginning) of the paragraph of the text sentence.

Furthermore, if there is a table of contents in the text, the position corresponding to the page described in the table of contents can be used as the table of contents information. Further, in this case, the titles (titles indicating the contents of chapters or sections) also described in the table of contents can be used as the table of contents information.

After creating the table of contents information, the process proceeds to step S5,
The text sentence is morphologically analyzed. That is, the language processing unit 1
Stores a dictionary (not shown), and in step S5,
By referring to word information and grammatical rules registered (stored) in the dictionary in advance, the text sentence is morphologically analyzed to be decomposed into words, and the part of speech is determined. As a result, for example, when a text sentence as shown in FIG. 4A is input, as the morphological analysis result, text information as a set of a word and its part of speech as shown in FIG. 4B is output. To be done.

Then, in step S6, a method for extracting a word (specific word) as a keyword from the text sentence is selected.

Here, the language processing unit 1 can extract a specific word by an extraction method such as a noun extraction method, a high-occurrence-frequency word extraction method, a low-occurrence-frequency word extraction method, or a designated word extraction method. Has been done.

In step S6, for example, the above-mentioned four methods are displayed on the liquid crystal display 18 of the input device 6 (FIG. 1) together with the cursor, and the user can move the jog shuttle 11 (shuttle ring 11a). Alternatively, operate the jog dial 11b) to move the cursor to indicate the desired method, and then press the select button 1
By operating 2, the extraction method is selected.

In this case, the specific word extraction method selection signal corresponding to the selected extraction method is output from the input device 6 to the language processing unit 1, and the process proceeds to step S7 to select the specific word extraction method from the input device 6. With the extraction method corresponding to the signal,
A specific word is extracted from the text sentence (morpheme analysis result).

In this case, according to the noun extraction method, a word whose part of speech is a noun is extracted as a specific word from the morphological analysis result (text information) in step S5.

According to the high appearance frequency word extraction method or the low appearance frequency extraction method, the number of the same independent words in the morphological analysis result of step S5 is counted, and the count value is, for example, a predetermined threshold value or more. Large or small words are extracted as specific words. In this case, it is also possible to extract a predetermined number of words as specific words in the descending order of the count value.

Further, according to the designated word extracting method, the keyboard 17 of the input device 6 is operated to input a word (one or a plurality of words) designated by the user, and this word is designated as a specific word. To be extracted.

The specific word extracted in step S7 is provided with, for example, a predetermined mark to distinguish it from other words. That is, the specific word information is added to the text information that is the result of morphological analysis.

In step S2 as well as in step S3, not only the chapter and section boundaries of the text sentence but also the paragraph boundaries are detected (it can be easily detected based on the structure information). In step S2 or S3, the boundary portion of the chapter, section, or paragraph of the text sentence can be marked with a mark indicating the boundary.

Further, in step S4, the user can operate the input device 6 so that the portion desired by the user as the table of contents can be used as the table of contents information. Further, by operating the input device 6 (keyboard 17), the title desired by the user can be added to the table of contents information.

The table-of-contents information and the marked text information (morpheme analysis result) obtained as described above are output to the storage unit 2 and stored as text information with reading control information.

Next, the voice synthesis by the text information with the reading control information stored in the storage unit 2 as described above, that is, the reading of the text sentence is performed as shown in the flowchart of FIG. That is, first, in step S11, the reading mode is selected.

Here, as the reading mode, for example, a mode of reading the entire sentence from the beginning of the text sentence, a mode of starting the reading from the portion where the previous reading was interrupted, and table of contents information of the text sentence (of chapters and sections) There is a form in which only the subject is read). Further, the table of contents information is displayed on the liquid crystal display 18, and the user is allowed to select a chapter, section, or paragraph, and reading is started from the selected portion. A few sentences from the beginning of the chapter, section, or paragraph. There is also a form in which only the sentences are read sequentially or in a predetermined order, a form in which only a specific word is read, and a sentence in which only a specific word or a random access point described later is read.

There is also a mode in which the order desired by the user is set at the beginning of the chapter, section, or paragraph based on the table of contents information, and the chapter, section, or paragraph is read aloud in that order.
In this case, a specific word or a random access point can be included in the order setting target. When this form is selected, the reading is performed in the order desired by the user.

Further, for example, when it is desired to know the outline of a text sentence, it is possible to make the text sentence read aloud so to speak by selecting the form in which only the table of contents information or only specific words are read. The contents can be grasped in a short time.

In step S11, the above-mentioned reading mode is displayed on the liquid crystal display 18 of the input device 6 together with, for example, the cursor, and the user operates the jog shuttle 11 (rotation operation), for example. And move the cursor to indicate the desired form,
The reading mode is selected by operating the selection button 12.

Then, in step S12, it is determined whether or not there is information that needs to be set (hereinafter referred to as reading required information) for the reading style selected in step S11. In step S12, for example, a question message asking (whether to ask) whether or not it is necessary to set the required reading information is displayed on the liquid crystal display 18, while the user jog shuttles the message. 11 and the selection button 12 or the keyboard 17 are operated to select whether or not to set necessary reading information. Then, in step S12, the above-described determination is performed corresponding to the selection result.

When it is determined in step S12 that there is no necessary reading information to be set, step S13 is skipped and the process proceeds to step S14. Also, step S
If it is determined in 12 that there is recitation necessary information to be set, the process proceeds to step S13, and recitation necessary information is set. That is, in step S13, for example, it is instructed whether reading is performed in the direction from the beginning to the end (hereinafter, referred to as forward direction) of the text sentence or in the direction from the end to the beginning (hereinafter referred to as backward direction). Or
Settings are made to instruct the start position of reading. Specifically, for example, after a mode in which only a specific word is read aloud is selected in step S11, a setting is made in step S13 to read a specific word in a text sentence forward or backward.

This reading device emits a predetermined sound (for example, a buzzer sound) at the boundary of a chapter, a section, or a paragraph during the reading of a text sentence so that the boundary of the chapter, the section, or the paragraph is changed. The user can be notified, and whether or not to emit this sound is also set in step S13.
It is performed in the process of setting the required reading information.

Further, in step S13, necessary reading information is displayed on the liquid crystal display 18 as necessary (for example, according to the selection result of the reading mode in step S11), and the liquid crystal is also displayed. Operate the jog shuttle 11 to move the cursor displayed on the display 18 to indicate the desired reading necessary information,
By manipulating and confirming the selection button 12, necessary reading information can be set.

The reading style selected in step S11,
The required reading information set in step S13 is output from the input device 6 to the reading control unit 3, and the reading control unit 3 waits for the operation of the reading button 16 of the input device 6 to obtain the reading mode and the reading required information. Therefore, the reading of the text information from the storage unit 2 is performed with reference to the reading control information stored in the storage unit 2 as necessary (however,
The necessary reading information is used only when it is set in step S13).

That is, when the reading button 16 is operated after it is determined in step S12 that there is no reading required information or after the reading information is set in step S13, the process proceeds to step S14, and the reading control unit 3, the text information is read from the storage unit 2 while referring to the recitation control information according to the recitation mode and the recitation required information (as described above, the recitation required information is set only in step S13). It is output and supplied to the generation unit 4.

As described above, the generation unit 4 generates phonological information (reading) and prosody information corresponding to the text information from the reading control unit 3 and outputs it to the acoustic processing unit 5.

When the text information is, for example, a sentence, the generation unit 4 generates phonological information and prosody information as follows, thereby generating a synthetic voice that is closer to human speech. , And is output from the sound processing unit 5.

That is, the generation unit 4 refers to a built-in dictionary (not shown), adds reading (pronunciation) to each word constituting the text information as shown in FIG. Performs processing such as vowel lengthening, vowel devoicing, and unvoiced consonant voicing. Further, in the generation unit 4, the dictionary is referred to, and the basic accent of reading the word (the word is
The accent when only one of them is spoken) is determined.

FIG. 6 (a) shows the result obtained by adding reading and basic accents to the morphological analysis result (text information) of FIG. 4 (b). In addition, in FIG. 6A, the reading is shown in Roman letters. Also, the symbol "'",
"@", "$", "X", and "Q" mean the position of accent, devoicing, consonant, vowel, and nasal noise, respectively.
For example, in the second parenthesis (the person's ordinary noun HO'XNIX) on the top line in FIG. 6A, the word "person" has a part of speech "ordinary noun" and a reading "HO'XNI".
X "(really) and its accent is" HO "
It is shown in (ho).

Further, in the generation unit 4, the morpheme analysis result (text information) is subjected to the bunsetsu identification process and the bunsetsu relation extraction process, and further, as prosody information, phrase information and accent position of the sentence. , Phoneme power,
The phoneme duration, the position and length of the pose, etc. are determined with reference to the dictionary.

That is, first, a phrase is identified from the morphological analysis result (text information) shown in FIG. 4B, for example. Specifically, the words are connected based on their part of speech, so that the phrase of the text information shown in FIG.
It is identified as shown in (b).

The results of identifying the bunsetsu are grouped into bunsetsu based on the relationship between the bunsetsus, for example, as shown in FIG. 6 (c) (in FIG. 6 (c), the collected bunsetsus are enclosed in parentheses ( ) Is shown in brackets). Then, the position and length of the pose inserted between the clauses are set. For example, in FIG. 6C, the pose may be inserted immediately after the closing bracket among the positions outside the bracket shown in FIG. 6, so the position of the pose is, for example, the number of mora. It is decided based on.

FIG. 6 (d) shows the result obtained by setting the pause for the text information (FIG. 6 (c)) in which the clauses are collected. In addition, in FIG.
Pose is represented by slash "/".

Further, the generating unit 4 obtains the structure of the sentence from the relation between the clauses of the text information (FIG. 6 (d)) for which the pose has been set, and based on this structure of the sentence, the sentence structure A phrase (phrase information) is set. It should be noted that the phrase information, when the voice synthesis in the acoustic processing unit 5,
For example, it is used to determine a pitch pattern (pitch pattern of voice unit data).

Then, the generation unit 4 uses the text information (FIG. 6 (d)) in which phrases and the like are set, and the text information (FIG. 6 (a)) in which the above-mentioned reading is added.
The phoneme information and the prosody information as shown in FIG. 6E are generated and output to the acoustic processing unit 5. Here, in FIG. 6E, the pose is indicated by "PAU".

At this time, the generation unit 4 can be made to determine the position of the accent in the phrase based on the accent movement rule of the phrase or the compound word. That is, when words are connected to form a sentence, the accent position of the word may change depending on whether the word is uttered in the sentence or when the word is uttered independently. It is possible to perform a so-called correction corresponding to the change in the position.

The acoustic processing section 5 performs speech synthesis based on the phoneme information and the prosody information from the generation section 4 and outputs a synthesized sound. As described above, step S14 (FIG. 5)
Then, if the text sentence has the reading style selected in step S11 and the reading necessary information is set in step S13, the text sentence is read according to the reading necessary information.

When reading is started as described above,
In step S15, it is determined whether to set a random access point, that is, whether the access point button 13 of the input device 6 has been operated by the user. When it is determined in step S15 that the access point button 13 has been operated, the process proceeds to step S16, a point for random access (random access point) is set, and the process proceeds to step S17.

That is, when the access point button 13 is operated, a control signal indicating that the access point button 13 has been operated is output from the input device 6 to the reading control information adding section 7, and the reading control information adding section 7 When this control signal is received, of the text information stored in the storage unit 2, a mark indicating a random access point is displayed at the portion (position) that is read aloud when the access point button 13 is operated. The content stored in the storage unit 2 is updated by being added as reading control information.

The position where the random access point is set is a reading position where the user desires to hear the synthesized sound again, for example.
When the mode in which only the sentence including the random access point is read in 1 is selected, the reading control unit 3
Only the text information corresponding to the sentence including the random access point will be read from the storage unit 2. This allows the user to immediately hear only the desired portion.

On the other hand, if it is determined in step S15 that the access point button 13 has not been operated, then step S16 is skipped and the process proceeds to step S17 to determine whether or not to set the table of contents information, that is, the input device 6 It is determined whether or not the index button 14 has been operated by the user.

In step S17, the contents button 14
When it is determined that is operated, the process proceeds to step S18, the table of contents information is set, and the process proceeds to step S19.

That is, when the table-of-contents button 14 is operated, a control signal indicating that the table-of-contents button 14 has been operated is output from the input device 6 to the reading control information giving section 7, and the reading control information giving section 7 outputs the control signal. When the control signal is received, of the text information stored in the storage unit 2, the portion (position) read aloud when the table-of-contents button 14 is operated,
It is newly added to the table of contents information and stored in the storage unit 2 (the table of contents information is updated).

On the other hand, when it is determined in step S17 that the table-of-contents button 14 has not been operated, step S18 is skipped, the process proceeds to step S19, and the reading mode selected in step S11 is further used in step S1.
If the reading necessary information is set in 3, whether or not the reading by reading the text information stored in the storage unit 2 is completed according to the reading necessary information, that is, the text information to be read is It is determined whether or not it is not stored in 2.

If it is determined in step S19 that the text information to be read is still stored in the storage unit 2, that is, if the reading has not been completed, the process proceeds to step S20 and the interruption button 15 of the input device 6 is entered. Is determined by the user. Step S2
When it is determined that the interruption button 15 has been operated at 0, the reading control unit 3 stops reading the text information from the storage unit 2, and thus the reading of the text sentence from the sound processing unit 5 is performed. Output is stopped, step S
Return to 11.

In this case, the recitation mode is selected again in step S11, and the recitation necessary information is reset (newly set) in step S13 if necessary. After that, as described above, when the reading button 16 is operated,
In step S14, the reading of the text sentence is resumed from the position being read aloud, for example, when the interruption button 15 is operated, according to the reselected reading style. In this case, the reading can be started from the beginning of the text sentence.

If it is determined in step S20 that the interruption button 15 has not been operated, the process returns to step S14 and the reading is continued.

On the other hand, in step S19, when it is determined that the text information to be read is not stored in the storage unit 2, the process ends.

Therefore, for example, the noun extraction method is selected as the method of extracting the specific word, and the mode of reading only the specific word is selected as the reading style, and the text sentence "the present invention is a book, newspaper, magazine, etc. Related to an automatic reading device that receives the type information of a digitized kanji-kana mixed sentence and listens with a synthetic voice. ”, First, in the language processing unit 1, the words“ present invention ”,“ book ”,
"Newspaper", "magazine", "digitization", "kanji and kana mixed sentences", "print information", "synthesized voice", "listening", and "automatic reading device" were extracted and marked. It is stored in the storage unit 2.

When the reading button 16 is operated,
Only the words marked with the specific word are read from the storage unit 2 by the reading control unit 3, and the “present invention”, “book”,
The reading is performed like "newspaper", "magazine", "electronic", "kanji / kana mixed sentence", "print information", "synthesized voice", "listening", "automatic reading device".

Note that the reading is performed in this way when the necessary reading information is an instruction to perform the reading in the forward direction, and the reading information is set to instruct the reading to be performed in the backward direction. If so, "automatic reading device",
The reading is performed like "listening", "synthesized voice", "print information", "kanji / kana mixed sentence", "electronic", "magazine", "newspaper", "book", "present invention".

In this case, since there are no chapters or sections in the text sentence, the table of contents information is not included in the language processing unit 1 unless the user operates the input device 6 to input the table of contents information. Not created.

Although the reading device to which the present invention is applied has been described above, the present invention can be applied not only to a kana-kana mixed sentence but also to a device for reading a sentence composed only of kana or an English sentence. it can.

In this embodiment, the information to be selected and set by the user is input by the jog shuttle 11 or the like operated by the user's hand. You can enter it.

Further, in the present embodiment, the information which the apparatus needs to inform the user is displayed on the liquid crystal display 18, but it is also possible to output the information from the sound processing section 5 as a synthetic sound. is there.

Further, in the present embodiment, the voice synthesis is performed by using the phoneme information and the prosody information, but it is also possible to perform the voice synthesis only by the phoneme information.

[0112]

As described above, according to the present invention, the content of a text sentence can be grasped in a short time.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a reading device to which the present invention has been applied.

FIG. 2 is a diagram showing a more detailed configuration of a user information input device 6 of FIG.

FIG. 3 is a flowchart illustrating an operation of the language processing unit 1 of FIG.

4 is a diagram showing a morpheme result by the language processing unit 1 of FIG.

FIG. 5 is a flowchart illustrating an operation when reading is performed according to the embodiment of FIG.

FIG. 6 is a diagram illustrating a process of a phoneme / prosodic information generation unit 4 of FIG.

FIG. 7 is a block diagram showing a configuration of an example of a conventional reading device.

[Explanation of symbols]

 1 language processing unit 2 text information storage unit with reading control information 3 reading control unit 4 phoneme / prosodic information generation unit 5 acoustic processing unit 6 user information input device 7 reading control information addition unit 11 jog shuttle 11a shuttle ring 11b jog dial 12 selection Button 13 Access point button 14 Table of contents button 15 Suspend button 16 Reading button 17 Keyboard 18 Liquid crystal display 21 Language processing section 22 Prosody processing section 23 Sound processing section

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁶ Identification number Internal reference number FI Technical indication location G10L 5/04 F (72) Inventor Koji Asano 6-735 Kitashinagawa, Shinagawa-ku, Tokyo Soni -Inside the corporation

Claims (25)

[Claims] 1. A reading information creating device for creating reading information for reading a text sentence, wherein the reading information includes at least information for obtaining phonological information of the text sentence and the text sentence. An aloud reading information generating device, characterized in that it includes aloud control information for reading. 2. The recitation information creating apparatus according to claim 1, wherein the recitation control information includes random access information indicating a predetermined position of the text sentence. 3. The recitation information creating apparatus according to claim 1, wherein the recitation control information includes table of contents information indicating a table of contents of the text sentence. 4. The table of contents information includes chapters of the text sentence,
The reading information creating device according to claim 3, wherein the reading information creating device is created based on a boundary of a section or a paragraph. 5. The paragraph of the text sentence is detected based on a line feed code in the text sentence, and the chapter or section boundary of the text sentence is determined by the number of characters, the number, or the chapter or section included in the paragraph. 5. The reading information creating apparatus according to claim 4, wherein the table of contents information is created based on a word representing the beginning part of the table and based on the boundary of the chapter or section. 6. The reading information creation according to claim 3, wherein when the text structure is provided with text structure information, the table of contents information is created based on the structure information. apparatus. 7. The reading information creating apparatus according to claim 3, further comprising a table of contents information input means for inputting the table of contents information. 8. The reading information creating apparatus according to claim 1, wherein the reading control information includes a specific word in the text sentence. 9. The reading according to claim 8, further comprising word input means for inputting a word, wherein the word input to the word input means is extracted from the text sentence as the specific word. Information creation device. 10. The reading information creating apparatus according to claim 8, wherein the text sentence is morphologically analyzed, and a word having a predetermined morpheme is extracted from the text sentence as the specific word. 11. A morphological analysis is performed on the text sentence, an appearance frequency of an independent word in the text sentence is obtained based on the analysis result, and a high occurrence frequency is extracted from the text sentence as the specific word. The reading information creating device according to claim 8. 12. A morphological analysis is performed on the text sentence, an appearance frequency of an independent word in the text sentence is calculated based on the analysis result, and a low occurrence frequency is extracted from the text sentence as the specific word. The reading information creating device according to claim 8. 13. A reading device for reading the text sentence based on the reading information of the text sentence, wherein the reading information reads at least information for obtaining phonological information of the text sentence and the text sentence. A reading device including reading control information for reading. 14. The reading control information includes random access information indicating a predetermined position of the text sentence, and the predetermined position of the text sentence is accessed based on the random access information. Item 1
The reading device according to item 3. 15. The reading device according to claim 13, further comprising a table of contents information input means for inputting table of contents information representing a table of contents of the text sentence as the reading control information. 16. The reading control information includes index information indicating an index of the text sentence.
The reading device according to item 3. 17. The reading device according to claim 16, wherein only the table-of-contents information is read aloud. 18. The reading device according to claim 16, wherein the table of contents information includes boundary information indicating boundaries of chapters, sections, or paragraphs of the text sentence. 19. The reading device according to claim 18, wherein only the beginning portion of the chapter, section, or paragraph of the text sentence is read based on the boundary information. 20. The reading device according to claim 18, wherein chapters, sections, or paragraphs of the text sentence are read aloud in a predetermined order based on the boundary information. 21. The reading device according to claim 18, wherein a boundary of a chapter, a section, or a paragraph of the text sentence is notified based on the boundary information during reading of the text sentence. 22. The reading control information includes a specific word in the text sentence, and reads only the specific word, or the specific word and a portion before or after the specific word. The reading device according to item 13. 23. Random access information input means for inputting random access information pointing to a predetermined position of the text sentence as the reading control information, and based on the random access information, to a predetermined position of the text sentence. The access is performed according to claim 1.
The reading device according to item 3. 24. The reading control information is random access information indicating a predetermined position in the text sentence, table of contents information indicating a table of contents of the text sentence, or a specific word in the text sentence. The reading device according to claim 13, wherein the reading is performed by accessing the text sentence based on the text sentence. 25. The reading device according to claim 13, wherein the reading of the text sentence is performed in the direction from the beginning to the end or from the end to the beginning.