JPH11344996A - Pronunciation document creating device, pronunciation document creating method and recording medium readable by computer in which program to make computer execute the method is recorded - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prepare and edit a pronunciation document including a pause, a metrical pattern based on converted accent, accent information and part of speech information by converting character information into the accent information and the part of speech information. SOLUTION: A pronunciation document creating device is provided with an input part 301 to input a character string, a converting part 302 to convert the inputted character string into the accent, the accent information and the part of speech information corresponding its reading, an accent phrase generating part 303 to generate an accent phrase based on the converted accent, the accent information and a part of speech, a pause setting part 304 to set pause information in what position between plural generated accent phrases, how long a silent section (pause) is inserted, etc., and a metrical pattern generating part 305 to generate the metrical patterns like a pitch pattern and time length of each syllable, etc., by the sentence to be constituted of the plural accent phrases in which the pause information is set.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pronunciation method for converting a character string constituting a document into an accent corresponding to the reading of each character string, and further creating a pronunciation document to which a pause and a prosodic pattern are added. The present invention relates to a document creation device, a pronunciation document creation method, and a computer-readable recording medium that records a program for causing a computer to execute the method.

[0002]

2. Description of the Related Art As one of information transmission and storage methods, a method using character information is known. As a method using this character information, in recent years, a document creator creates a desired document using a document creation device such as a Japanese word processor or an English word processor, or a personal computer having a word processor function, and transmits the created document to a network. Via the Internet, or a method of storing a created document in a recording medium such as a magnetic disk or an optical disk. This is due to the development of computer-related technology, which has realized higher functionality and lower cost of the document creation device itself, as well as the promotion of paperless offices, the development of communication networks, and the spread of e-mail. It depends heavily on environmental changes.

As other methods used for transmitting and storing information, a method using audio information and a method using audio information and video information are known. For example,
In the method using voice information, the voice information is directly transmitted via a telephone or the like, and the information is stored and recorded on a tape or the like using a recording device. In the method using audio information and video information, information is transmitted using a communication device having a monitor and a speaker, and the audio information and video information are transferred, and information is stored using a recording device such as a video device. They are stored on video tapes, optical disks, etc.

[0004] Among the above-mentioned methods of transmitting and storing information, the method using character information is different from other methods.
It is the most versatile and widely used because it has a small amount of data, is easy to edit information, and can be used as digital information on a computer.

[0005]

However, according to the above-described conventional method using character information, since the created document is information limited to visual linguistic information (that is, character linguistic information), non-linguistic information is not used. Expressions such as emotions could not be added as information. In the linguistic information using speech (that is, speech linguistic information), expressions such as emotions corresponding to non-linguistic information can be expressed by changing the way of speaking such as accent, loudness (volume), and pitch of voice. Was added as information.

Further, in the prior art, there has not been provided an apparatus and a method for creating information in which information of two expression forms, that is, character information and audio information, is combined in a consistent manner.

In general, audio information is basically edited by using the auditory sense (that is, by listening to the reproduced audio information with the ear). Since the (location) needs to be confirmed, the operation is complicated and troublesome.

It is possible to synthesize speech from a text document (that is, character information) by using a text speech synthesis technique which is one of the conventional speech synthesis techniques. Then, there was a problem that proper nouns not included in the dictionary were misread or pronounced with different accents. Further, there are problems that it is not possible to express emotions or the like which are non-verbal information, and it is not possible to synthesize speech accurately according to the way the creator intends to speak.

The present invention has been made in view of the above, and includes a pause and a prosodic pattern based on converted accents and accent / part of speech information by converting character information into accent information and part of speech information. The purpose is to be able to create and edit pronunciation documents.

[0010]

Means for Solving the Problems The above-mentioned problems are solved,
In order to achieve the object, a pronunciation document creation device according to claim 1 includes an input unit for inputting a character string, and converting the character string input by the input unit into accent and accent information / speech information corresponding to the reading. Conversion means for converting;
An accent phrase generating means for generating an accent phrase based on the accent and the accent information / part of speech converted by the converting means; and at what position and how long between the plurality of accent phrases generated by the accent phrase generating means. A pause setting means for setting pause information such as whether a silent section (pause) is inserted, and a pitch pattern / each syllable in units of sentences composed of a plurality of accent phrases in which the pause information is set by the pause setting means. And a prosody pattern generating means for generating a prosody pattern such as a time length.

According to the first aspect of the present invention, by inputting a character string and converting the character string at a delimiter position of an accent phrase, it is possible to accurately input in a manner intended by the document creator without inputting natural speech. A pronunciation document for synthesizing speech can be created, and the efficiency and convenience of creating a pronunciation document can be improved.

According to a second aspect of the present invention, in the pronunciation document creating apparatus according to the first aspect, the accent phrase generated by the accent phrase generating means and / or the pause information and / or the pause information set by the pause setting means are set. A display means for displaying a prosody pattern generated by the prosody pattern generation means is provided.

According to the second aspect of the present invention, since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the efficiency of creation of the pronunciation document and the convenience can be improved. .

According to a third aspect of the present invention, there is provided the pronunciation document creating apparatus according to the first or second aspect, further comprising: an accent phrase generated by the accent phrase generating means;
Voice synthesis means for synthesizing voice using the pause information set by the pause setting means and the prosody pattern generated by the prosody pattern generation means, and voice output for outputting the voice synthesized by the voice synthesis means Means,
It is characterized by having.

According to the third aspect of the present invention, since the pronunciation data of the pronunciation document is synthesized and output, the contents of the pronunciation document can be easily reproduced, and the efficiency of creation of the pronunciation document and the convenience are improved. It is possible to achieve.

According to a fourth aspect of the present invention, in the pronunciation document creating apparatus according to any one of the first to third aspects, the accent phrase generated by the accent phrase generating means and / or the pause setting means are set. Editing means for editing the generated pause information and / or the prosody pattern generated by the prosody pattern generation means.

According to the fourth aspect of the present invention, the pronunciation document can be edited, so that a pronunciation document closer to the way of speech intended by the creator can be easily created, and the efficiency and convenience of creating the pronunciation document can be improved. It is possible to improve the performance.

According to a fifth aspect of the present invention, in the pronunciation document creating method, a first step of inputting a character string and converting the character string input in the first step into accent and accent information / speech information corresponding to the reading. A second step of converting, and a third step of generating an accent phrase based on the accent and accent information converted by the second step and the part of speech.
And a fourth step of setting pose information such as a position and a length of a silent section (pause) between the plurality of accent phrases generated in the third step and at which position, A fifth step of generating a prosody pattern such as a pitch pattern and a time length of each syllable in a sentence unit composed of a plurality of accent phrases in which pause information is set in the step.

According to the fifth aspect of the present invention, by inputting a character string and converting the character string at a delimiter position of an accent phrase, it is possible to accurately input a character string intended by a document creator without inputting natural speech. A pronunciation document for synthesizing speech can be created, and the efficiency and convenience of creating a pronunciation document can be improved.

According to a sixth aspect of the present invention, there is provided the pronunciation document creating method according to the fifth aspect, further comprising the accent phrase generated in the third step and / or the pause information and / or the pause information set in the fourth step. Or the fifth
The method includes a sixth step of displaying the prosodic pattern generated by the step.

According to the sixth aspect of the present invention, since the pronunciation document is displayed, the content of the pronunciation document can be easily confirmed, and the efficiency of creation of the pronunciation document and the convenience can be improved. .

According to a seventh aspect of the present invention, there is provided the pronunciation document creating method according to the fifth or sixth aspect, further comprising: the accent phrase generated in the third step; and the pose information set in the fourth step. , A seventh step of synthesizing speech using the prosodic pattern generated in the fifth step, and an eighth step of outputting the speech synthesized in the seventh step.

According to the seventh aspect of the present invention, since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the efficiency of creation of the pronunciation document and the convenience can be improved. .

According to an eighth aspect of the present invention, there is provided the pronunciation document creating apparatus according to any one of the fifth to seventh aspects, further comprising: an accent phrase generated in the third step and / or a setting in the fourth step. A ninth step of editing the generated pause information and / or the prosodic pattern generated in the fifth step.

According to the eighth aspect of the present invention, since the pronunciation document can be edited, it is possible to easily create a pronunciation document that is closer to the way of speech intended by the creator, and the efficiency and convenience of creating the pronunciation document are improved. It is possible to improve the performance.

Further, a storage medium according to a ninth aspect of the present invention is:
By recording a program for causing a computer to execute the method according to any one of claims 5 to 8, the program becomes machine-readable, whereby
The operation of any one of claims 5 to 8 can be realized by a computer.

[0027]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a pronunciation document creation apparatus, a pronunciation document creation method, and a computer-readable recording medium that records a program for causing a computer to execute the method according to the present invention will be described below in detail. .

(Embodiment 1) FIG. 1 is a block diagram showing a hardware configuration of a pronunciation document creating apparatus 100 according to Embodiment 1. The pronunciation document creation device 100 includes a control unit 101, an application storage unit 102, a conversion dictionary 103, a prosody pattern model storage unit 104, a timbre storage unit 105, a key input device 106, and a display device 107.
, A microphone 108, a speaker 109, a pronunciation document storage unit 110, an interface (I / F) 111, a floppy disk drive (FD drive) 112,
It comprises a D-ROM drive 113 and a communication unit 114.

The control unit 101 is a central processing unit for controlling the above-mentioned units connected to the bus BS.
1a, a ROM 101b, a RAM 101c, and the like. The CPU 101a stores the O stored in the ROM 101b.
It operates according to an S (operating system) program or an application program stored in the application storage unit 102. ROM 101b
Is a memory for storing an OS program.
1c is a memory used as a work area for various programs.

The application storage unit 102 includes a reading accent conversion application that realizes a function of converting reading to accent, a pause setting application that realizes a function of setting a pause, and a prosody pattern that realizes a function of generating a prosody pattern. Various application programs such as a generation application are stored. Further, the pronunciation document creation device 100 according to the first embodiment includes:
It has a Kana-Kanji conversion function, and a Kana-Kanji conversion application that realizes the Kana-Kanji conversion function is also stored in the application storage unit 102.

The conversion dictionary 103 is a dictionary in which the reading of a character string (word) and the character string containing the corresponding kanji are stored in association with each other. It is also a database-based dictionary that stores information and part-of-speech information in association with each other.

The prosody pattern model storage unit 104 is a memory that stores a model of the prosody pattern in advance and is made into a database. The contents of the prosody pattern stored in the prosody pattern model storage unit 104 will be described later.

The timbre storage unit 105 stores selectable timbre data representing acoustic parameters for each unit such as accent for each timbre type. The voice storage unit 105
Voice data or the like can be added through a communication line, a storage medium such as the FD 112a, the CD-ROM 113a, or the like, or can be deleted by a key operation of the key input device 106.

The key input device 106 includes an input device such as a keyboard and a mouse, and is used to perform various operations such as input of a character string, designation of reproduction of a pronunciation document, and creation / registration of a pronunciation document. . Further, the key input device 106 includes a conversion key for converting an input character string into a character string containing kanji.

The display 107 is a liquid crystal display or C
An RT display is used for displaying character strings, displaying pronunciation documents, displaying various messages, and the like.

The microphone 108 is used to sample the original real voice which is the original voice waveform data used for creating a model of a prosody pattern.

The speaker 109 is used for reproducing and outputting the voice synthesized by the voice synthesizing unit 105 and reproducing various sounds.

The pronunciation document storage unit 110 is a memory for storing the created pronunciation document. As will be described in detail later, the pronunciation document is document data corresponding to an input character string including data related to accent phrases, data related to pauses, data related to prosodic patterns, and the like.

The I / F 111 is a unit for exchanging data between the bus BS and the FD drive 112 or the CD-ROM drive 113. The FD drive 112 carries out an operation of reading and writing data by mounting a removable FD 112a (recording medium). The CD-ROM drive 113 carries out an operation of reading data by mounting a removable CD-ROM 113a (recording medium). The pronunciation document stored in the pronunciation document storage unit 110 is stored in the F / F drive 112 via the I / F 111 and the FD drive 112.
It is also possible to save it in D112a.

The communication unit 114 is connected to a communication line and communicates with an external device via the communication line.

In the first embodiment, the key input device 1
A case of inputting a character string via the input line 06 will be described as an example. However, the present invention is not limited to this. A handwriting input device is connected, and a character string is input by determining (character recognition) a handwritten character. Alternatively, a character string may be input from a word processor document or the like created in advance.

FIG. 2 is an external view of the pronunciation document creating apparatus 100 according to the first embodiment. As illustrated, as a hardware configuration, a personal computer having a microphone 108 and a speaker 109 can be used.

Next, the configuration of the pronunciation document creating apparatus 100 according to the first embodiment will be functionally described. FIG. 3 is a functional block diagram illustrating a functional configuration of the pronunciation document creation device 100 according to the first embodiment. In FIG. 3, the pronunciation document creation device 10
0 indicates an input unit 301, a conversion unit 302, an accent phrase generation unit 303, a pause setting unit 304, a prosody pattern generation unit 305, a display unit 306, a speech synthesis unit 307,
The configuration includes an audio output unit 308.

The input section 301 is for inputting a kana character string. Specifically, for example, the key input device 106
Or by inputting kana characters through the communication unit 114.

The conversion unit 302 converts the kana character string input by the input unit 301 into a character string including a kanji corresponding to the reading, and also converts the character string into an accent corresponding to the reading. The part of speech information is added to the accent. The conversion process is specifically performed by, for example, pressing a not-shown conversion key provided on the key input device 106.

The conversion unit 302 converts the reading of the particles "wa" or "he" into accents "wa" or "e" based on the part of speech information. Also, the conversion unit 302
Also has a long reading or function. For example, if you enter "Gakuko", it will be converted to "Gacco" which is an accent closer to a more natural way of speaking.

The accent phrase generation unit 303 includes the input unit 30
Reading and conversion unit 30 for words in the character string input by
The accent phrase is generated by combining a plurality of accents based on the accent information and the part of speech converted by step 2. The accent information includes, for example, an accent type. Further, an accent combination attribute or the like may be included in the accent information.

FIG. 4 is an explanatory diagram showing the generation of accent phrases in the pronunciation document creation device according to the first embodiment. FIG.
In the input character string "kabushiki", the accent corresponding to the reading, "type 2 accent" as accent information, and "noun" as part of speech information are stored in the conversion dictionary. Operation, the conversion unit 3
With 02, "2 type accent" as accent information (accent type) and "noun" as part of speech information are added to the input character string "kabushiki". Similarly, the conversion unit 302 adds “0-type accent” as accent information (accent type) and “noun” as part-of-speech information to the input character string “gaisha”. Similarly, the conversion unit 302 does not particularly add accent information to the input character string “no”, but adds only “particles” as part of speech information.

The accent phrase generator 303 combines the converted accents based on the accent information and the part-of-speech information, thereby converting the input character strings "kabushiki", "gaisha", "no" to "kabushiga'ishana". Is generated.

The pause setting section 304 sets pose information such as where and how long a silent section (pause) is to be inserted between a plurality of accent phrases generated by the accent phrase generation section 303. It is. The setting of the pose information, that is, the insertion of the pose, is performed manually. Specifically, for example, it is realized by inserting a pause of a desired length at a desired position of the accent phrase displayed by the display device 107. In addition to the manual insertion of the pose, it is also possible to automatically insert the pose according to a predetermined condition in a fixed pose or the like. The inserted pose is also displayed on the display device 107.

The prosody pattern generation unit 305 generates a prosody pattern such as a pitch pattern and a time length of each syllable in units of sentences composed of a plurality of accent phrases in which the pause information is set by the pause setting unit 304. is there. Specifically, a prosody pattern is generated by obtaining a pitch pattern and a time length of each syllable for each sentence.

As a method of generating a prosody pattern, a method using a plurality of types of prosody pattern models stored in a prepared prosody pattern model storage unit 104 can be considered. A prosody pattern can be generated by extracting an optimal model from among the plurality of types of prosody pattern models from conditions such as accent information, accent positions, and the number of accent phrases. Further, the prosody pattern may be such that the operator selects a desired model and uses the desired model. Furthermore, a prosody pattern may be generated by selecting from prosody patterns in a database using the above conditions.

The display unit 306 displays a pronunciation document, that is, an accent phrase generated by the accent phrase generation unit 303 and / or pause information set by the pause setting unit 304 and / or a prosody pattern generation unit 305.
The prosody pattern generated by the means is displayed. Specifically, for example, it is realized by using the display device 107. The display unit 306
May display the reading of the kanji in ruby at the kanji portion of the character string containing the converted kanji.

Further, the speech synthesis unit 307 generates the accent phrase generated by the accent phrase generation unit 303, the pause information set by the pause setting unit 304, and the pronunciation composed of the prosody pattern generated by the prosody pattern generation unit 305. The voice is synthesized using the document and the timbre data stored in the timbre storage unit 105. The audio output unit 308 outputs the audio synthesized by the audio synthesis unit 307. Specifically, it is realized by the speaker 109 and the like.

The input unit 301, the conversion unit 302, the accent phrase generation unit 303, the pause setting unit 304, the prosody pattern generation unit 305, the display unit 306, the voice synthesis unit 307, and the voice output unit 308 are respectively ROM101b and RAM1.
01c or the application storage unit 102, the floppy disk 112a, the CD-ROM 113a, and the like.
The function of each part is realized.

In the above configuration, a procedure of a series of processes for creating a pronunciation document will be described. FIG. 5 shows the first embodiment.
4 is a flowchart showing the procedure of a pronunciation document creation process of the pronunciation document creation device of FIG. In the flowchart of FIG.
First, the process waits for a character input from the key input device 106 (step S501). Input of characters is performed in the same manner as in a normal word processor.

When a character is input (step S5)
(01 affirmative), it is determined whether or not the conversion key has been pressed (step S502). The conversion key is usually pressed at the time of a phrase break or at the time of an accent phrase break. Here, if the conversion key is not pressed (No at step S502), it is determined that there is still a character input, not a break of a clause or the like, and the process proceeds to step S501 to wait for further character input.

When the conversion key is pressed (step S5)
(02 affirmative), it is determined whether or not an accent has been input from the key input device 106 or the like (step S50).
3) If there is a direct input of an accent (step S
(503 affirmative) shifts to step S506 without doing anything.

If there is no input of an accent at step S503 (No at step S503), the accent and the accent information / speech information corresponding to the input character string are read out from the conversion dictionary 103 and converted into accents. Then, accent information and part of speech information are added (step S504). At this time, if a plurality of accents, accent information / speech information candidates exist for the character string in the conversion dictionary 103, the candidates are displayed by the same method as the conventional kana-kanji conversion method, By making the selection, the conversion can be determined (step S505). However, it is determined whether or not the conversion has been determined (step S505). If the conversion has not been determined (No at step S505), the process proceeds to step S504, and the conversion process is repeated until the conversion is determined.

If the conversion is determined (Yes at step S505), then, based on the reading of the input character string, the accent converted at step S504, and the accent information / part-of-speech information added to the accent, A phrase is generated (step S506). The method of generating the accent phrase is as described above. After that, the procedure moves to step S507.

In step S507, the conversion dictionary 103 is used to convert the character string input in step S501 into a character string containing kanji. The conversion method is the same as the conventional Kana-Kanji conversion method. It is determined whether or not the conversion is determined (step S508), and if the conversion is not determined (No at step S508), step S50 is performed.
7 and the conversion process (step S507) is repeated until the conversion is determined. When the conversion is determined (Yes at Step S508), the process proceeds to Step S509.

It is to be noted that the kana-kanji conversion is performed after conversion into accents and addition of accent information and part-of-speech information, but the order may be reversed. If there are conversion candidates, character string candidates including kanji and accents, accent information / speech information candidates are simultaneously displayed, and a desired combination is selected from among them to confirm the conversion. Is also good.

Next, in step S509, it is determined whether or not there is an input of a punctuation mark indicating a sentence delimiter or a return key. If there is no input (step S509: No), the process proceeds to step S501, and the process proceeds to step S501.
Steps up to S509 are repeated.

In step S509, if there is an input of a period or a return key (Yes in step S509), a pause is set (step S509).
510). The method of setting the pose is as described above.
In this processing step, discrete prosody information can be obtained.

Next, a prosody pattern is generated (step S511). The method of generating the prosody pattern is as described above. Further, the pronunciation document in which the prosody pattern has been generated is stored in the pronunciation document storage unit 110 (step S512).

Further, the pronunciation document is displayed on the display device 107 (step S513), and all the processing ends.

Next, the sound output processing of the pronunciation document will be described. FIG. 6 is a flowchart illustrating a procedure of a sound output process of a pronunciation document in the pronunciation document creation device according to the first embodiment. In the flowchart of FIG.
It is determined whether a voice output instruction has been given (step S6).
01), and waits for an output instruction (Yes at step S601).
The corresponding pronunciation document is read from the pronunciation document storage unit 110 (step S602).

Next, speech synthesis is performed using the read pronunciation data of the pronunciation document and the timbre data stored in the timbre storage unit 105 (step S603). afterwards,
The synthesized voice is output using the speaker 109 (step S604).

As described above, according to the first embodiment, it is possible to create a pronunciation document simply by inputting a character string and pressing a conversion key in the same manner as when performing kana-kanji conversion. At this time, by using the timing information of the conversion key of the Kana-Kanji conversion as an accent phrase delimiter position, errors in accent phrase delimiter can be reduced. In addition, the created pronunciation document can be displayed. Furthermore, the contents of the created pronunciation document can be synthesized and output.

In the first embodiment, only the method of creating a Japanese pronunciation document has been described. However, the present invention is not limited to this, and an English pronunciation document creation method may be used. In that case, if you input the spelling of the word and convert it to an accent corresponding to the reading (pronunciation) of the word in response to the input of the space key between the words, you can create Can create English pronunciation documents.

(Embodiment 2) In Embodiment 1 described above, a pronunciation string is created by inputting a character string and pressing a conversion key as in the case of performing kana-kanji conversion. However, as in the second embodiment described below, the pronunciation document may be edited in order to change the once created pronunciation document to a more natural way of speaking.

For the hardware configuration and external view of the pronunciation document generation device 700 according to the second embodiment of the present invention, refer to FIG. 1 and FIG.
Since it is the same as 00, its description is omitted. FIG. 7 is a functional block diagram showing a functional configuration of the pronunciation document creation device 700. In FIG. 7, the pronunciation document creation device 7
00 is the same as in the first embodiment except for the editing unit 700.
3 has the same configuration as that of each part of the pronunciation document creating apparatus 100 shown in FIG.

In FIG. 7, the editing unit 701 includes an accent phrase generated by the accent phrase generation unit 303 and / or pause information set by the pause setting unit 304 and / or a prosody pattern generated by the prosody pattern generation unit 305. Is to edit. Specifically, correction of reading, correction of accent type, insertion / deletion of a break of an accent phrase, change of pause information, change of prosody pattern, and the like are performed.

Since the reading is also corrected as described above, a part of the function of the conversion unit 302 capable of performing kanji and accent conversion corresponding to the reading is also provided.

The editing method is performed by inputting variable data using the key input device 106 with reference to the pronunciation document displayed on the display device 107. This is performed by the same method as editing a document created by a word processor.

Next, the procedure of the editing process will be described. FIG. 8 shows the editing unit 70 of the document creation device according to the second embodiment.
9 is a flowchart illustrating a procedure of one editing process. FIG.
First, it is determined whether or not there is an editing instruction (step S801), and after waiting for the editing instruction (Yes at step S801), the corresponding pronunciation document is read from the pronunciation document storage unit 110 (step S802).

Next, the read pronunciation document is edited (step S803). Thereafter, it is determined whether or not the editing process has been completed (step S804). If the editing process has not been completed (step S804: No), the process proceeds to step S80.
3 and the editing process is repeated.

If the editing process has been completed in step S804 (Yes at step S804), the pronunciation data of the edited pronunciation document and the timbre storage unit are checked in order to confirm the contents of the edited pronunciation document. Voice synthesis is performed using the timbre data stored in the memory 105 (step S805). Then, the synthesized voice is sent to the speaker 1
09 is output (step S806).

After the synthesized speech is confirmed, if the editing is to be performed again (Yes at Step S807), Step S80 is performed.
Then, the process proceeds to step S3, and the processes in steps S803 to S807 are repeatedly performed. When the editing is not performed again, that is, when the content of the editing is determined (No at Step S807), the pronunciation document subjected to the editing process is written in the pronunciation document storage unit 110 (Step S808), and all the processing ends. .

As described above, according to the second embodiment of the present invention, a pronunciation document that has been once created is edited, so that a pronunciation document that can be synthesized in a more natural way of speaking can be obtained. .

[0081]

As described above, according to the first aspect of the present invention, the input means for inputting a character string, and the character string input by the input means are converted to the accent and accent information / part of speech corresponding to the reading. Conversion means for converting the information into information, accent phrase generation means for generating an accent phrase based on the accent and the accent information and part of speech converted by the conversion means, and a plurality of accent phrases generated by the accent phrase generation means. Pose setting means for setting pause information such as at which position and how long a silence section (pause) is to be inserted, and a plurality of accent phrases in which the pause information is set by the pause setting means. Prosody pattern generation means for generating a prosody pattern such as a pitch pattern and a time length of each syllable in sentence units;
By inputting a character string and converting it at the delimiter position of the accent phrase, it is possible to create a pronunciation document for synthesizing speech accurately in the way that the document creator intends without inputting natural speech. It is possible to obtain a pronunciation document creation device that can create the pronunciation document and can improve the efficiency and convenience of creating the pronunciation document.

Further, according to the invention of claim 2, according to claim 1
The invention according to the invention, further comprising display means for displaying the accent phrase generated by the accent phrase generation means and / or the pause information set by the pause setting means and / or the prosody pattern generated by the prosody pattern generation means. Since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the pronunciation document creation device capable of improving the efficiency and convenience of creating the pronunciation document can be obtained.

Further, according to the invention of claim 3, according to claim 1
Alternatively, in the invention according to the second aspect, further, a speech is generated by using the accent phrase generated by the accent phrase generation unit, the pose information set by the pause setting unit, and the prosody pattern generated by the prosody pattern generation unit. Since there is provided a voice synthesizing means for synthesizing and a voice output means for outputting a voice synthesized by the voice synthesizing means, the pronunciation data of the pronunciation document is synthesized and output, so that the contents of the pronunciation document can be easily reproduced This makes it possible to obtain a pronunciation document creation device capable of improving pronunciation document creation efficiency and convenience.

According to the invention of claim 4, according to claim 1,
In the invention according to any one of the first to third aspects, the accent phrase and / or
Alternatively, since there is provided editing means for editing the pause information set by the pause setting means and / or the prosody pattern generated by the prosody pattern generation means, the pronunciation document can be edited, so that the creator's intention You can easily create pronunciation documents closer to how you speak,
There is an effect that a pronunciation document creation device capable of improving the efficiency and convenience of creating pronunciation documents can be obtained.

According to the fifth aspect of the present invention, the first step of inputting a character string and converting the character string input in the first step into accent and accent information / speech information corresponding to the reading. A second step, a third step of generating an accent phrase based on the accent and accent information and the part of speech converted in the second step, and a position between the plurality of accent phrases generated in the third step. A fourth step of setting pause information such as how long a silence section (pause) is to be inserted in the fourth step;
A fifth step of generating a prosody pattern such as a pitch pattern and a time length of each syllable in a sentence unit composed of a plurality of accent phrases in which pose information is set by the step. By converting at the phrase delimiter, it is possible to create a pronunciation document for synthesizing speech accurately according to the way the creator intends without inputting natural speech, and to create a pronunciation document There is an effect that a pronunciation document creation method capable of improving efficiency and convenience can be obtained.

Further, according to the invention of claim 6, according to claim 5,
And a sixth step of displaying the accent phrase generated in the third step and / or the pause information set in the fourth step and / or the prosodic pattern generated in the fifth step. Therefore, since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the pronunciation document creation method capable of improving the efficiency and convenience of creating the pronunciation document is obtained.

According to the invention of claim 7, according to claim 5,
Alternatively, in the invention according to the sixth aspect, speech is synthesized using the accent phrase generated in the third step, the pause information set in the fourth step, and the prosodic pattern generated in the fifth step. Since the method includes the seventh step and the eighth step of outputting the voice synthesized in the seventh step, the pronunciation document is displayed, so that the content of the pronunciation document can be easily confirmed, and the creation of the pronunciation document can be performed. There is an effect that a pronunciation document creation method capable of improving efficiency and convenience can be obtained.

The invention according to claim 8 is the invention according to any one of claims 5 to 7, further comprising the accent phrase generated in the third step and / or the pose information set in the fourth step. And / or the fifth
Since the method includes a ninth step of editing the prosodic pattern generated by the step, the pronunciation document can be edited, so that a pronunciation document closer to the way of speech intended by the creator can be easily created, and the pronunciation document can be easily created. There is an effect that a pronunciation document creation method capable of improving the creation efficiency and convenience of the pronunciation document can be obtained.

The storage medium according to the ninth aspect of the present invention
By recording a program for causing a computer to execute the method according to any one of claims 5 to 8, the program becomes machine-readable, whereby
An advantage is obtained in that a recording medium capable of realizing the operation of any one of claims 5 to 8 by a computer is obtained.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of a pronunciation document creation device according to a first embodiment of the present invention.

FIG. 2 is an external view of a pronunciation document creation device according to the first embodiment.

FIG. 3 is a functional block diagram illustrating a functional configuration of a pronunciation document creation device according to the first embodiment;

FIG. 4 is an explanatory diagram showing generation of an accent phrase in the pronunciation document creation device according to the first embodiment;

FIG. 5 is a flowchart illustrating a procedure of a pronunciation document creation process in the pronunciation document creation device according to the first embodiment;

FIG. 6 is a flowchart illustrating a procedure of a sound output process in the pronunciation document creation device according to the first embodiment;

FIG. 7 is a functional block diagram illustrating a functional configuration of a pronunciation document creation device according to a second embodiment of the present invention;

FIG. 8 is a flowchart illustrating a procedure of editing a pronunciation document in the pronunciation document creation device according to the second embodiment;

[Explanation of symbols]

 Reference Signs List 100, 700 pronunciation document creation device 101 control unit 101a CPU 101b ROM 101c RAM 102 application storage unit 103 conversion dictionary 104 prosody pattern model storage unit 105 voice color storage unit 106 key input device 107 display device 108 microphone 109 speaker 110 pronunciation document storage unit 111 Interface (I / F) 112 FD drive 113 CD-ROM drive 114 Communication unit 301 Input unit 302 Conversion unit 303 Accent phrase generation unit 304 Pause setting unit 305 Prosody pattern generation unit 306 Display unit 307 Voice synthesis unit 308 Audio output unit 701 Editing Department

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] July 12, 1999

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Full text

[Correction method] Change

[Correction contents]

[Document Name] Statement

Patent application title: Pronunciation document creation device, pronunciation document creation method, and computer-readable recording medium recording a program for causing a computer to execute the method

[Claims]

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pronunciation method for converting a character string constituting a document into an accent corresponding to the reading of each character string, and further creating a pronunciation document to which a pause and a prosodic pattern are added. The present invention relates to a document creation device, a pronunciation document creation method, and a computer-readable recording medium that records a program for causing a computer to execute the method.

[0002]

2. Description of the Related Art As one of information transmission and storage methods, a method using character information is known. As a method using this character information, in recent years, a document creator creates a desired document using a document creation device such as a Japanese word processor or an English word processor, or a personal computer having a word processor function, and transmits the created document to a network. Via the Internet, or a method of storing a created document in a recording medium such as a magnetic disk or an optical disk. This is due to the development of computer-related technology, which has realized higher functionality and lower cost of the document creation device itself, as well as the promotion of paperless offices, the development of communication networks, and the spread of e-mail. It depends heavily on environmental changes.

As other methods used for transmitting and storing information, a method using audio information and a method using audio information and video information are known. For example,
In the method using voice information, the voice information is directly transmitted via a telephone or the like, and the information is stored and recorded on a tape or the like using a recording device. In the method using audio information and video information, information is transmitted using a communication device having a monitor and a speaker, and the audio information and video information are transferred, and information is stored using a recording device such as a video device. They are stored on video tapes, optical disks, etc.

[0004] Among the above-mentioned methods of transmitting and storing information, the method using character information is different from other methods.
It is the most versatile and widely used because it has a small amount of data, is easy to edit information, and can be used as digital information on a computer.

[0005]

However, according to the above-described conventional method using character information, since the created document is information limited to visual linguistic information (that is, character linguistic information), non-linguistic information is not used. Expressions such as emotions could not be added as information. In the linguistic information using speech (that is, speech linguistic information), expressions such as emotions corresponding to non-linguistic information can be expressed by changing the way of speaking such as accent, loudness (volume), and pitch of voice. Was added as information.

Further, in the prior art, there has not been provided an apparatus and a method for creating information in which information of two expression forms, that is, character information and audio information, is combined in a consistent manner.

In general, audio information is basically edited by using the auditory sense (that is, by listening to the reproduced audio information with the ear). Since the (location) needs to be confirmed, the operation is complicated and troublesome.

It is possible to synthesize speech from a text document (that is, character information) by using a text speech synthesis technique which is one of the conventional speech synthesis techniques. Then, there was a problem that proper nouns not included in the dictionary were misread or pronounced with different accents. Further, there are problems that it is not possible to express emotions or the like which are non-verbal information, and it is not possible to synthesize speech accurately according to the way the creator intends to speak.

The present invention has been made in view of the above, and includes a pause and a prosodic pattern based on converted accents and accent / part of speech information by converting character information into accent information and part of speech information. The purpose is to be able to create and edit pronunciation documents.

[0010]

Means for Solving the Problems The above-mentioned problems are solved,
In order to achieve the object, a pronunciation document creation device according to claim 1 comprises: input means for inputting a kana character string; and converting the kana character string input by the input means into a character string containing kanji corresponding to the reading. Conversion means for converting the accent into an accent corresponding to the reading, and at this time, a conversion means for adding the accent information and the part of speech information to the accent, and an accent based on the accent and the accent information / part of speech converted by the conversion means. Accent phrase generation means for generating a phrase, and pose information such as where and how long a silent section (pause) is to be inserted between the plurality of accent phrases generated by the accent phrase generation means are set. A pause setting unit, and a sentence unit composed of a plurality of accent phrases in which the pause information is set by the pause setting unit. Prosodic pattern generating means for generating a Tchipatan-prosodic pattern time length of each syllable, and further comprising a.

According to the first aspect of the present invention, by inputting a character string and converting the character string at a delimiter position of an accent phrase, it is possible to accurately input in a manner intended by the document creator without inputting natural speech. A pronunciation document for synthesizing speech can be created, and the efficiency and convenience of creating a pronunciation document can be improved.

According to a second aspect of the present invention, in the pronunciation document creating apparatus according to the first aspect, the accent phrase generated by the accent phrase generating means and / or the pause information and / or the pause information set by the pause setting means are set. A display means for displaying a prosody pattern generated by the prosody pattern generation means is provided.

According to the second aspect of the present invention, since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the efficiency of creation of the pronunciation document and the convenience can be improved. .

According to a third aspect of the present invention, there is provided the pronunciation document creating apparatus according to the first or second aspect, further comprising: an accent phrase generated by the accent phrase generating means;
Voice synthesis means for synthesizing voice using the pause information set by the pause setting means and the prosody pattern generated by the prosody pattern generation means, and voice output for outputting the voice synthesized by the voice synthesis means Means,
It is characterized by having.

According to the third aspect of the present invention, since the pronunciation data of the pronunciation document is synthesized and output, the contents of the pronunciation document can be easily reproduced, and the efficiency of creation of the pronunciation document and the convenience are improved. It is possible to achieve.

According to a fourth aspect of the present invention, in the pronunciation document creating apparatus according to any one of the first to third aspects, the accent phrase generated by the accent phrase generating means and / or the pause setting means are set. Editing means for editing the generated pause information and / or the prosody pattern generated by the prosody pattern generation means.

According to the fourth aspect of the present invention, the pronunciation document can be edited, so that a pronunciation document closer to the way of speech intended by the creator can be easily created, and the efficiency and convenience of creating the pronunciation document can be improved. It is possible to improve the performance.

According to a fifth aspect of the present invention, there is provided a pronunciation document creation method, wherein a first step of inputting a kana character string and a kana character string input in the first step are converted into a character string containing a kanji corresponding to the reading. A second step of converting the accent into an accent corresponding to the reading, and adding accent information and part-of-speech information to the accent.
A third step of generating an accent phrase based on the accent and accent information converted by the step and the part of speech, and a silence section at which position and how long between the plurality of accent phrases generated by the third step A fourth step of setting pause information such as whether to insert a (pause), and a pitch pattern, a time length of each syllable, etc. in units of sentences composed of a plurality of accent phrases in which the pause information is set in the fourth step. And a fifth step of generating a prosodic pattern of

According to the fifth aspect of the present invention, by inputting a character string and converting the character string at a delimiter position of an accent phrase, it is possible to accurately input a character string intended by a document creator without inputting natural speech. A pronunciation document for synthesizing speech can be created, and the efficiency and convenience of creating a pronunciation document can be improved.

According to a sixth aspect of the present invention, there is provided the pronunciation document creating method according to the fifth aspect, further comprising the accent phrase generated in the third step and / or the pause information and / or the pause information set in the fourth step. Or the fifth
The method includes a sixth step of displaying the prosodic pattern generated by the step.

According to the sixth aspect of the present invention, since the pronunciation document is displayed, the content of the pronunciation document can be easily confirmed, and the efficiency of creation of the pronunciation document and the convenience can be improved. .

According to a seventh aspect of the present invention, there is provided the pronunciation document creating method according to the fifth or sixth aspect, further comprising: the accent phrase generated in the third step; and the pose information set in the fourth step. , A seventh step of synthesizing speech using the prosodic pattern generated in the fifth step, and an eighth step of outputting the speech synthesized in the seventh step.

According to the seventh aspect of the present invention, since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the efficiency of creation of the pronunciation document and the convenience can be improved. .

According to an eighth aspect of the present invention, there is provided the pronunciation document creating apparatus according to any one of the fifth to seventh aspects, further comprising: an accent phrase generated in the third step and / or a setting in the fourth step. A ninth step of editing the generated pause information and / or the prosodic pattern generated in the fifth step.

According to the eighth aspect of the present invention, since the pronunciation document can be edited, it is possible to easily create a pronunciation document that is closer to the way of speech intended by the creator, and the efficiency and convenience of creating the pronunciation document are improved. It is possible to improve the performance.

Further, a storage medium according to a ninth aspect of the present invention is:
By recording a program for causing a computer to execute the method according to any one of claims 5 to 8, the program becomes machine-readable, whereby
The operation of any one of claims 5 to 8 can be realized by a computer.

[0027]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a pronunciation document creation apparatus, a pronunciation document creation method, and a computer-readable recording medium that records a program for causing a computer to execute the method according to the present invention will be described below in detail. .

(Embodiment 1) FIG. 1 is a block diagram showing a hardware configuration of a pronunciation document creating apparatus 100 according to Embodiment 1. The pronunciation document creation device 100 includes a control unit 101, an application storage unit 102, a conversion dictionary 103, a prosody pattern model storage unit 104, a timbre storage unit 105, a key input device 106, and a display device 107.
, A microphone 108, a speaker 109, a pronunciation document storage unit 110, an interface (I / F) 111, a floppy disk drive (FD drive) 112,
It comprises a D-ROM drive 113 and a communication unit 114.

The control unit 101 is a central processing unit for controlling the above-mentioned units connected to the bus BS.
1a, a ROM 101b, a RAM 101c, and the like. The CPU 101a stores the O stored in the ROM 101b.
It operates according to an S (operating system) program or an application program stored in the application storage unit 102. ROM 101b
Is a memory for storing an OS program.
1c is a memory used as a work area for various programs.

The application storage unit 102 includes a reading accent conversion application that realizes a function of converting reading to accent, a pause setting application that realizes a function of setting a pause, and a prosody pattern that realizes a function of generating a prosody pattern. Various application programs such as a generation application are stored. Further, the pronunciation document creation device 100 according to the first embodiment includes:
It has a Kana-Kanji conversion function, and a Kana-Kanji conversion application that realizes the Kana-Kanji conversion function is also stored in the application storage unit 102.

The conversion dictionary 103 is a dictionary in which the reading of a character string (word) and the character string containing the corresponding kanji are stored in association with each other. It is also a database-based dictionary that stores information and part-of-speech information in association with each other.

The prosody pattern model storage unit 104 is a memory that stores a model of the prosody pattern in advance and is made into a database. The contents of the prosody pattern stored in the prosody pattern model storage unit 104 will be described later.

The timbre storage unit 105 stores selectable timbre data representing acoustic parameters for each unit such as accent for each timbre type. The voice storage unit 105
Voice data or the like can be added through a communication line, a storage medium such as the FD 112a, the CD-ROM 113a, or the like, or can be deleted by a key operation of the key input device 106.

The key input device 106 includes an input device such as a keyboard and a mouse, and is used to perform various operations such as input of a character string, designation of reproduction of a pronunciation document, and creation / registration of a pronunciation document. . Further, the key input device 106 includes a conversion key for converting an input character string into a character string containing kanji.

The display 107 is a liquid crystal display or C
An RT display is used for displaying character strings, displaying pronunciation documents, displaying various messages, and the like.

The microphone 108 is used to sample the original real voice which is the original voice waveform data used for creating a model of a prosody pattern.

The speaker 109 is used for reproducing and outputting the voice synthesized by the voice synthesizing unit 105 and reproducing various sounds.

The pronunciation document storage unit 110 is a memory for storing the created pronunciation document. As will be described in detail later, the pronunciation document is document data corresponding to an input character string including data related to accent phrases, data related to pauses, data related to prosodic patterns, and the like.

The I / F 111 is a unit for exchanging data between the bus BS and the FD drive 112 or the CD-ROM drive 113. The FD drive 112 carries out an operation of reading and writing data by mounting a removable FD 112a (recording medium). The CD-ROM drive 113 carries out an operation of reading data by mounting a removable CD-ROM 113a (recording medium). The pronunciation document stored in the pronunciation document storage unit 110 is stored in the F / F drive 112 via the I / F 111 and the FD drive 112.
It is also possible to save it in D112a.

The communication unit 114 is connected to a communication line and communicates with an external device via the communication line.

In the first embodiment, the key input device 1
A case of inputting a character string via the input line 06 will be described as an example. However, the present invention is not limited to this. A handwriting input device is connected, and a character string is input by determining (character recognition) a handwritten character. Alternatively, a character string may be input from a word processor document or the like created in advance.

FIG. 2 is an external view of the pronunciation document creating apparatus 100 according to the first embodiment. As illustrated, as a hardware configuration, a personal computer having a microphone 108 and a speaker 109 can be used.

Next, the configuration of the pronunciation document creating apparatus 100 according to the first embodiment will be functionally described. FIG. 3 is a functional block diagram illustrating a functional configuration of the pronunciation document creation device 100 according to the first embodiment. In FIG. 3, the pronunciation document creation device 10
0 indicates an input unit 301, a conversion unit 302, an accent phrase generation unit 303, a pause setting unit 304, a prosody pattern generation unit 305, a display unit 306, a speech synthesis unit 307,
The configuration includes an audio output unit 308.

The input section 301 is for inputting a kana character string. Specifically, for example, the key input device 106
Or by inputting kana characters through the communication unit 114.

The conversion unit 302 converts the kana character string input by the input unit 301 into a character string including a kanji corresponding to the reading, and also converts the character string into an accent corresponding to the reading. The part of speech information is added to the accent. The conversion process is specifically performed by, for example, pressing a not-shown conversion key provided on the key input device 106.

The conversion unit 302 converts the reading of the particles "wa" or "he" into accents "wa" or "e" based on the part of speech information. Also, the conversion unit 302
Also has a function to lengthen the reading. For example, if you enter "Gakuko", it will be converted to "Gacco" which is an accent closer to a more natural way of speaking.

The accent phrase generation unit 303 includes the input unit 30
Reading and conversion unit 30 for words in the character string input by
The accent phrase is generated by combining a plurality of accents based on the accent information and the part of speech converted by step 2. The accent information includes, for example, an accent type. Further, an accent combination attribute or the like may be included in the accent information.

FIG. 4 is an explanatory diagram showing the generation of accent phrases in the pronunciation document creation device according to the first embodiment. FIG.
In the input character string "kabushiki", the accent corresponding to the reading, "type 2 accent" as accent information, and "noun" as part of speech information are stored in the conversion dictionary. Operation, the conversion unit 3
With 02, "2 type accent" as accent information (accent type) and "noun" as part of speech information are added to the input character string "kabushiki". Similarly, the conversion unit 302 adds “0-type accent” as accent information (accent type) and “noun” as part-of-speech information to the input character string “gaisha”. Similarly, the conversion unit 302 does not particularly add accent information to the input character string “no”, but adds only “particles” as part of speech information.

The accent phrase generator 303 combines the converted accents based on the accent information and the part-of-speech information, thereby converting the input character strings "kabushiki", "gaisha", "no" to "kabushiga'ishana". Is generated.

The pause setting section 304 sets pose information such as where and how long a silent section (pause) is to be inserted between a plurality of accent phrases generated by the accent phrase generation section 303. It is. The setting of the pose information, that is, the insertion of the pose, is performed manually. Specifically, for example, it is realized by inserting a pause of a desired length at a desired position of the accent phrase displayed by the display device 107. In addition to the manual insertion of the pose, it is also possible to automatically insert the pose according to a predetermined condition in a fixed pose or the like. The inserted pose is also displayed on the display device 107.

The prosody pattern generation unit 305 generates a prosody pattern such as a pitch pattern and a time length of each syllable in units of sentences composed of a plurality of accent phrases in which the pause information is set by the pause setting unit 304. is there. Specifically, a prosody pattern is generated by obtaining a pitch pattern and a time length of each syllable for each sentence.

As a method of generating a prosody pattern, a method using a plurality of types of prosody pattern models stored in a prepared prosody pattern model storage unit 104 can be considered. A prosody pattern can be generated by extracting an optimal model from among the plurality of types of prosody pattern models from conditions such as accent information, accent positions, and the number of accent phrases. Further, the prosody pattern may be such that the operator selects a desired model and uses the desired model. Furthermore, a prosody pattern may be generated by selecting from prosody patterns in a database using the above conditions.

The display unit 306 displays a pronunciation document, that is, an accent phrase generated by the accent phrase generation unit 303 and / or pause information set by the pause setting unit 304 and / or a prosody pattern generation unit 305.
The prosody pattern generated by the means is displayed. Specifically, for example, it is realized by using the display device 107. The display unit 306
May display the reading of the kanji in ruby at the kanji portion of the character string containing the converted kanji.

Further, the speech synthesis unit 307 generates the accent phrase generated by the accent phrase generation unit 303, the pause information set by the pause setting unit 304, and the pronunciation composed of the prosody pattern generated by the prosody pattern generation unit 305. The voice is synthesized using the document and the timbre data stored in the timbre storage unit 105. The audio output unit 308 outputs the audio synthesized by the audio synthesis unit 307. Specifically, it is realized by the speaker 109 and the like.

The input unit 301, the conversion unit 302, the accent phrase generation unit 303, the pause setting unit 304, the prosody pattern generation unit 305, the display unit 306, the voice synthesis unit 307, and the voice output unit 308 are respectively ROM101b and RAM1.
01c or the application storage unit 102, the floppy disk 112a, the CD-ROM 113a, and the like.
The function of each part is realized.

In the above configuration, a procedure of a series of processes for creating a pronunciation document will be described. FIG. 5 shows the first embodiment.
4 is a flowchart showing the procedure of a pronunciation document creation process of the pronunciation document creation device of FIG. In the flowchart of FIG.
First, the process waits for a character input from the key input device 106 (step S501). Input of characters is performed in the same manner as in a normal word processor.

When a character is input (step S5)
(01 affirmative), it is determined whether or not the conversion key has been pressed (step S502). The conversion key is usually pressed at the time of a phrase break or at the time of an accent phrase break. Here, if the conversion key is not pressed (No at step S502), it is determined that there is still a character input, not a break of a clause or the like, and the process proceeds to step S501 to wait for further character input.

When the conversion key is pressed (step S5)
(02 affirmative), it is determined whether or not an accent has been input from the key input device 106 or the like (step S50).
3) If there is a direct input of an accent (step S
(503 affirmative) shifts to step S506 without doing anything.

If there is no input of an accent at step S503 (No at step S503), the accent and the accent information / speech information corresponding to the input character string are read out from the conversion dictionary 103 and converted into accents. Then, accent information and part of speech information are added (step S504). At this time, if a plurality of accents, accent information / speech information candidates exist for the character string in the conversion dictionary 103, the candidates are displayed by the same method as the conventional kana-kanji conversion method, By making the selection, the conversion can be determined (step S505). Therefore, it is determined whether or not the conversion is determined (step S505). If the conversion is not determined (No at step S505), the process proceeds to step S504, and the conversion process is repeated until the conversion is determined.

If the conversion is determined (Yes at step S505), then, based on the reading of the input character string, the accent converted at step S504, and the accent information / part-of-speech information added to the accent, A phrase is generated (step S506). The method of generating the accent phrase is as described above. After that, the procedure moves to step S507.

In step S507, the conversion dictionary 103 is used to convert the character string input in step S501 into a character string containing kanji. The conversion method is the same as the conventional Kana-Kanji conversion method. It is determined whether or not the conversion is determined (step S508), and if the conversion is not determined (No at step S508), step S50 is performed.
7 and the conversion process (step S507) is repeated until the conversion is determined. When the conversion is determined (Yes at Step S508), the process proceeds to Step S509.

It is to be noted that the kana-kanji conversion is performed after conversion into accents and addition of accent information and part-of-speech information, but the order may be reversed. If there are conversion candidates, character string candidates including kanji and accents, accent information / speech information candidates are simultaneously displayed, and a desired combination is selected from among them to confirm the conversion. Is also good.

Next, in step S509, it is determined whether or not there is an input of a punctuation mark indicating a sentence delimiter or a return key. If there is no input (step S509: No), the process proceeds to step S501, and the process proceeds to step S501.
Steps up to S509 are repeated.

In step S509, if there is an input of a period or a return key (Yes in step S509), a pause is set (step S509).
510). The method of setting the pose is as described above.
In this processing step, discrete prosody information can be obtained.

Next, a prosody pattern is generated (step S511). The method of generating the prosody pattern is as described above. Further, the pronunciation document in which the prosody pattern has been generated is stored in the pronunciation document storage unit 110 (step S512).

Further, the pronunciation document is displayed on the display device 107 (step S513), and all the processing ends.

Next, the sound output processing of the pronunciation document will be described. FIG. 6 is a flowchart illustrating a procedure of a sound output process of a pronunciation document in the pronunciation document creation device according to the first embodiment. In the flowchart of FIG.
It is determined whether a voice output instruction has been given (step S6).
01), and waits for an output instruction (Yes at step S601).
The corresponding pronunciation document is read from the pronunciation document storage unit 110 (step S602).

Next, speech synthesis is performed using the read pronunciation data of the pronunciation document and the timbre data stored in the timbre storage unit 105 (step S603). afterwards,
The synthesized voice is output using the speaker 109 (step S604).

As described above, according to the first embodiment, it is possible to create a pronunciation document simply by inputting a character string and pressing a conversion key in the same manner as when performing kana-kanji conversion. At this time, by using the timing information of the conversion key of the Kana-Kanji conversion as an accent phrase delimiter position, errors in accent phrase delimiter can be reduced. In addition, the created pronunciation document can be displayed. Furthermore, the contents of the created pronunciation document can be synthesized and output.

In the first embodiment, only the method of creating a Japanese pronunciation document has been described. However, the present invention is not limited to this, and an English pronunciation document creation method may be used. In that case, if you input the spelling of the word and convert it to an accent corresponding to the reading (pronunciation) of the word in response to the input of the space key between the words, you can create Can create English pronunciation documents.

(Embodiment 2) In Embodiment 1 described above, a pronunciation string is created by inputting a character string and pressing a conversion key as in the case of performing kana-kanji conversion. However, as in the second embodiment described below, the pronunciation document may be edited in order to change the once created pronunciation document to a more natural way of speaking.

For the hardware configuration and external view of the pronunciation document generation device 700 according to the second embodiment of the present invention, refer to FIG. 1 and FIG.
Since it is the same as 00, its description is omitted. FIG. 7 is a functional block diagram showing a functional configuration of the pronunciation document creation device 700. In FIG. 7, the pronunciation document creation device 7
00 is the same as in the first embodiment except for the editing unit 700.
3 has the same configuration as that of each part of the pronunciation document creating apparatus 100 shown in FIG.

In FIG. 7, the editing unit 701 includes an accent phrase generated by the accent phrase generation unit 303 and / or pause information set by the pause setting unit 304 and / or a prosody pattern generated by the prosody pattern generation unit 305. Is to edit. Specifically, correction of reading, correction of accent type, insertion / deletion of a break of an accent phrase, change of pause information, change of prosody pattern, and the like are performed.

Since the reading is also corrected as described above, a part of the function of the conversion unit 302 capable of performing kanji and accent conversion corresponding to the reading is also provided.

The editing method is performed by inputting variable data using the key input device 106 with reference to the pronunciation document displayed on the display device 107. This is performed by the same method as editing a document created by a word processor.

Next, the procedure of the editing process will be described. FIG. 8 shows the editing unit 70 of the document creation device according to the second embodiment.
9 is a flowchart illustrating a procedure of one editing process. FIG.
First, it is determined whether or not there is an editing instruction (step S801), and after waiting for the editing instruction (Yes at step S801), the corresponding pronunciation document is read from the pronunciation document storage unit 110 (step S802).

Next, the read pronunciation document is edited (step S803). Thereafter, it is determined whether or not the editing process has been completed (step S804). If the editing process has not been completed (step S804: No), the process proceeds to step S80.
3 and the editing process is repeated.

If the editing process has been completed in step S804 (Yes at step S804), the pronunciation data of the edited pronunciation document and the timbre storage unit are checked in order to confirm the contents of the edited pronunciation document. Voice synthesis is performed using the timbre data stored in the memory 105 (step S805). Then, the synthesized voice is sent to the speaker 1
09 is output (step S806).

After the synthesized speech is confirmed, if the editing is to be performed again (Yes at Step S807), Step S80 is performed.
Then, the process proceeds to step S3, and the processes in steps S803 to S807 are repeatedly performed. When the editing is not performed again, that is, when the content of the editing is determined (No at Step S807), the pronunciation document subjected to the editing process is written in the pronunciation document storage unit 110 (Step S808), and all the processing ends. .

As described above, according to the second embodiment of the present invention, a pronunciation document that has been once created is edited, so that a pronunciation document that can be synthesized in a more natural way of speaking can be obtained. .

[0081]

As described above, according to the first aspect of the present invention, an input means for inputting a kana character string and a character including a kanji corresponding to the reading of the kana character string input by the input means are provided. In addition to the conversion into a column, the conversion into an accent corresponding to the reading, the conversion means for adding accent information and part of speech information to the accent, and the accent and accent information / part of speech converted by the conversion means An accent phrase generating means for generating an accent phrase, and pose information such as where and how long a silent section (pause) is to be inserted between the plurality of accent phrases generated by the accent phrase generating means. A sentence unit composed of a pause setting unit to be set and a plurality of accent phrases for which pose information has been set by the pause setting unit Prosody pattern generation means for generating a prosody pattern such as a pitch pattern and a time length of each syllable, so that a character string is input and converted at an accent phrase delimiter without inputting natural speech. Thus, there is provided a pronunciation document creation apparatus capable of creating a pronunciation document for accurately synthesizing speech in a manner intended by the document creator and improving the efficiency and convenience of pronunciation document creation. The effect is that it can be done.

Further, according to the invention of claim 2, according to claim 1
The invention according to the invention, further comprising display means for displaying the accent phrase generated by the accent phrase generation means and / or the pause information set by the pause setting means and / or the prosody pattern generated by the prosody pattern generation means. Since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the pronunciation document creation device capable of improving the efficiency and convenience of creating the pronunciation document can be obtained.

Further, according to the invention of claim 3, according to claim 1
Alternatively, in the invention according to the second aspect, further, a speech is generated by using the accent phrase generated by the accent phrase generation unit, the pose information set by the pause setting unit, and the prosody pattern generated by the prosody pattern generation unit. Since there is provided a voice synthesizing means for synthesizing and a voice output means for outputting a voice synthesized by the voice synthesizing means, the pronunciation data of the pronunciation document is synthesized and output, so that the contents of the pronunciation document can be easily reproduced This makes it possible to obtain a pronunciation document creation device capable of improving pronunciation document creation efficiency and convenience.

According to the invention of claim 4, according to claim 1,
In the invention according to any one of the first to third aspects, the accent phrase and / or
Alternatively, since there is provided editing means for editing the pause information set by the pause setting means and / or the prosody pattern generated by the prosody pattern generation means, the pronunciation document can be edited, so that the creator's intention You can easily create pronunciation documents closer to how you speak,
There is an effect that a pronunciation document creation device capable of improving the efficiency and convenience of creating pronunciation documents can be obtained.

According to the fifth aspect of the present invention, the first step of inputting a kana character string and the kana character string input in the first step are converted into a character string containing a kanji corresponding to the reading. And a second step of converting the accent into an accent corresponding to the reading, and adding the accent information and the part of speech information to the accent, and an accent based on the accent, the accent information and the part of speech converted in the second step. A third step of generating a phrase, and a fourth step of setting pause information such as where and how long a silence section (pause) is to be inserted between the plurality of accent phrases generated in the third step. And generating a prosodic pattern such as a pitch pattern and a time length of each syllable in units of sentences composed of a plurality of accent phrases for which pause information is set in the fourth step. The fifth step includes inputting a character string and converting the character string at a delimiter position of an accent phrase, thereby accurately synthesizing a speech according to a speech style intended by a document creator without inputting a natural speech. Therefore, there is an effect that a pronunciation document creation method that can improve the efficiency and convenience of creating pronunciation documents can be obtained.

Further, according to the invention of claim 6, according to claim 5,
And a sixth step of displaying the accent phrase generated in the third step and / or the pause information set in the fourth step and / or the prosodic pattern generated in the fifth step. Therefore, since the pronunciation document is displayed, the contents of the pronunciation document can be easily confirmed, and the pronunciation document creation method capable of improving the efficiency and convenience of creating the pronunciation document is obtained.

According to the invention of claim 7, according to claim 5,
Alternatively, in the invention according to the sixth aspect, speech is synthesized using the accent phrase generated in the third step, the pause information set in the fourth step, and the prosodic pattern generated in the fifth step. Since the method includes the seventh step and the eighth step of outputting the voice synthesized in the seventh step, the pronunciation document is displayed, so that the content of the pronunciation document can be easily confirmed, and the creation of the pronunciation document can be performed. There is an effect that a pronunciation document creation method capable of improving efficiency and convenience can be obtained.

The invention according to claim 8 is the invention according to any one of claims 5 to 7, further comprising the accent phrase generated in the third step and / or the pose information set in the fourth step. And / or the fifth
Since the method includes a ninth step of editing the prosodic pattern generated by the step, the pronunciation document can be edited, so that a pronunciation document closer to the way of speech intended by the creator can be easily created, and the pronunciation document can be easily created. There is an effect that a pronunciation document creation method capable of improving the creation efficiency and convenience of the pronunciation document can be obtained.

The storage medium according to the ninth aspect of the present invention
By recording a program for causing a computer to execute the method according to any one of claims 5 to 8, the program becomes machine-readable, whereby
An advantage is obtained in that a recording medium capable of realizing the operation of any one of claims 5 to 8 by a computer is obtained.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of a pronunciation document creation device according to a first embodiment of the present invention.

FIG. 2 is an external view of a pronunciation document creation device according to the first embodiment.

FIG. 3 is a functional block diagram illustrating a functional configuration of a pronunciation document creation device according to the first embodiment;

FIG. 4 is an explanatory diagram showing generation of an accent phrase in the pronunciation document creation device according to the first embodiment;

FIG. 5 is a flowchart illustrating a procedure of a pronunciation document creation process in the pronunciation document creation device according to the first embodiment;

FIG. 6 is a flowchart illustrating a procedure of a sound output process in the pronunciation document creation device according to the first embodiment;

FIG. 7 is a functional block diagram illustrating a functional configuration of a pronunciation document creation device according to a second embodiment of the present invention;

FIG. 8 is a flowchart illustrating a procedure of editing a pronunciation document in the pronunciation document creation device according to the second embodiment;

[Description of Signs] 100, 700 Pronunciation document creation device 101 Control unit 101a CPU 101b ROM 101c RAM 102 Application storage unit 103 Conversion dictionary 104 Prosody pattern model storage unit 105 Voice storage unit 106 Key input device 107 Display device 108 Microphone 109 Speaker 110 Pronunciation document storage unit 111 Interface (I / F) 112 FD drive 113 CD-ROM drive 114 Communication unit 301 Input unit 302 Conversion unit 303 Accent phrase generation unit 304 Pause setting unit 305 Prosodic pattern generation unit 306 Display unit 307 Voice synthesis unit 308 Voice output unit 701 Editing unit

Claims (9)

[Claims] An input unit for inputting a character string; a conversion unit for converting the character string input by the input unit into accent and accent information / part of speech information corresponding to the reading; An accent phrase generating means for generating an accent phrase based on the accent and the accent information / part of speech; And a pause setting unit for setting pause information such as whether to insert a pitch pattern.
A prosody pattern generating means for generating a prosody pattern such as a time length of each syllable. 2. A display means for displaying the accent phrase generated by the accent phrase generation means and / or the pause information set by the pause setting means and / or the prosody pattern generated by the prosody pattern generation means. The pronunciation document creating apparatus according to claim 1, further comprising: 3. A speech synthesizer using the accent phrase generated by the accent phrase generation unit, the pause information set by the pause setting unit, and the prosody pattern generated by the prosody pattern generation unit. 3. The pronunciation document creating device according to claim 1, further comprising: a voice synthesizing unit that performs voice synthesis; 4. An editing means for editing the accent phrase generated by the accent phrase generating means and / or the pause information set by the pause setting means and / or the prosodic pattern generated by the prosodic pattern generating means. 4. The method according to claim 1, wherein
The pronunciation document creation device according to any one of the above. 5. A first step of inputting a character string, a second step of converting the character string input in the first step into accent and accent information / speech information corresponding to the reading, and the second step A third step of generating an accent phrase based on the accent and accent information and the part-of-speech that have been converted by the first and second steps; A fourth step of setting pose information such as whether to insert a pause), and a pitch pattern, a time length of each syllable, and the like in a sentence unit composed of a plurality of accent phrases in which the pause information is set in the fourth step. A fifth step of generating a prosodic pattern; and a pronunciation document creating method. 6. A sixth step of displaying the accent phrase generated in the third step and / or the pause information set in the fourth step and / or the prosodic pattern generated in the fifth step. 6. The pronunciation document creation method according to claim 5, wherein the pronunciation document is included. 7. A method of synthesizing speech using the accent phrase generated in the third step, the pause information set in the fourth step, and the prosodic pattern generated in the fifth step. 7. The pronunciation document creation method according to claim 5, further comprising: a seventh step; and an eighth step of outputting the voice synthesized in the seventh step. A ninth step of editing the accent phrase generated in the third step and / or the pause information set in the fourth step and / or the prosodic pattern generated in the fifth step. The pronunciation document creation method according to any one of claims 5 to 7, further comprising: 9. A computer-readable recording medium on which a program for causing a computer to execute the method according to claim 5 is recorded.