JPH09160583A - Method and device for processing voice information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a variety of utterances by indicating the rhythm data of the inputted character-string, and uttering the character-string in accordance with the corrected rhythm data. SOLUTION: Data are received from an input part to a sentence input part 101, a corrected information label input part 110, and a used information label input part 114. The rhythm data kept in a rhythm data storage part 105 are indicated on a display screen of a display part by a rhythm data indicating part 107. In correcting the rhythm, the correction value is inputted by dragging a ○ with a mouse watching the display screen. When correction of the rhythm data is completed, the label information is inputted from the corrected information label input part 110 to the corrected data, and the inputted corrected information label is kept in the corrected information storage part 112 together with the rhythm data. In addition, a sound generating part 117 performs the sound generation treatment such as rhythm parameter connection, and the waveform connection referring to the values stored in a rhythm data correction and storage part 109.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice information processing method and apparatus, and more particularly to a voice information processing method and apparatus capable of editing prosodic data.

The present invention relates to a voice information processing method and device, and more particularly, to a voice information processing method and device that enables utterance of the same character with a plurality of patterns of intonation.

[0003]

2. Description of the Related Art Generally, a speech synthesizer analyzes an input sentence and further represents information such as accents and pauses by texts and frames (hereinafter referred to as phonetic data).
A prosody control unit for generating a pitch pattern by referring to the phonetic data received from the voice language processing unit, a prosody and a parameter connection referring to the pitch pattern received from the prosody control unit, It is divided into a voice generation unit that performs voice generation processing such as waveform connection and produces a synthetic voice via a DA converter.

An example of phonetic data for an input sentence "Yesterday was fine weather" is shown in FIG. 4, and an example of pitch patterns is shown in FIG.

However, the above-mentioned conventional device has a drawback in that it is possible to generate only one kind of synthetic speech having the highest likelihood for an input sentence having the same face.

[0006] That is, even in the case where it is desired to forcibly make a plain-text (inclination) intonation into a question tone (increase) or to make a standard-language intonation a Kansai dialect for sentences of the same character, Can't generate synthetic speech.

[0007]

SUMMARY OF THE INVENTION An object of the present invention is to eliminate the above-mentioned drawbacks, edit prosody data generated by a prosody control unit, and use the result to perform speech generation.
It is to realize various vocalizations. Further, it is to realize reuse of the edited value by giving a label to the edited result and holding / reading it.

[0008]

In order to solve the above problems, the present invention displays prosodic data of an input character string, corrects the displayed prosodic data, and according to the corrected prosodic data. Say a string.

In order to solve the above-mentioned problems, the present invention processes the input character string by speech language to generate the prosody data to be displayed.

In order to solve the above problems, the present invention, when displaying the prosody data, displays the phonetic notation corresponding to the prosody data in association with the prosody data.

In order to solve the above problem, the present invention corrects the prosody data by moving a mark representing the prosody data on the display screen.

In order to solve the above problems, the present invention provides that the character string is a character string specified in a displayed sentence.

In order to solve the above problems, the present invention adds a label to the modified prosody data and registers it.

In order to solve the above problems, the present invention inputs a label, reads prosody data corresponding to the input label, and utters according to the read prosody data.

In order to solve the above-mentioned problems, the present invention holds the modified prosody data, and executes utterance according to the held prosody data in response to an instruction of utterance.

In order to solve the above-mentioned problems, the present invention controls display means for displaying the prosody data.

In order to solve the above problems, the present invention controls so that the character string is uttered by a speaker.

In order to solve the above problems, the present invention processes modified data of the prosody data input from an input means.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram showing a schematic configuration of a speech synthesizer according to an embodiment of the present invention. In the figure, 1
Is a CPU, which performs various controls in the speech synthesizer. 2 is a ROM (Read Only Memory), C
The control program executed by the PU 1 and parameters used for various processes are stored. The ROM 2 also stores a control program for controlling the CPU 1 for executing the processing described in the flowchart described below. Reference numeral 3 denotes a RAM (random access memory), which provides an information storage unit as a work area for the CPU 1 to execute various controls. Reference numeral 4 is a voice output unit, which, based on the synthesized voice signal,
Generates and outputs audio.

Reference numeral 5 denotes an external memory which stores modification information of prosody data for voice synthesis processing. An input unit 6 such as a keyboard and a mouse is used to input a sentence to be voice-synthesized and to input correction information of prosody data. 7
Is a display unit, which is composed of a CRT, a liquid crystal display, etc.,
Display prosodic data etc. Reference numeral 8 denotes a bus, which connects the above-mentioned components to each other and enables data exchange between the components.

FIG. 2 is a block diagram showing the functional arrangement of the speech synthesizer of this embodiment.

First, the correspondence between the functional configurations shown in FIG. 2 and the configuration shown in FIG. 1 will be described.

The sentence input unit 101 and the correction information label input unit 1
10 and the usage information label input unit 114 input each data by the input unit 6.

Input sentence holding unit 102, phonetic data holding unit 1
04, the prosody data holding unit 106, the prosody data correction value holding unit 109, the correction information label holding unit 111, the use information label holding unit 115, and the voice parameter holding unit 118, R
Each parameter is stored in AM3.

The modification information holding unit 112 stores parameters of modification information in the external memory 5.

Speech language processing unit 103, prosody control unit 10
5, prosody data correction unit 108, correction information holding execution unit 11
3, the usage information read execution unit 116, the utterance execution unit 120, and the voice generation unit 117 are executed by the CPU 1.

The prosody data display unit 107 displays the prosody data on the display unit 7, and the DA converter 119 displays the voice output unit 4.
Run in

In FIG. 2, reference numeral 101 denotes a sentence input unit for inputting a sentence for speech synthesis from the input unit 6. An input sentence holding unit (buffer) 102 holds the input sentence input from the sentence input unit 101 in the RAM 3.

Reference numeral 103 is a voice language processing unit for performing voice language processing on the input sentence to generate phonetic data such as voice notation, accent type, pause length, utterance speed, and volume, and 104 is a voice language. It is a phonetic sound data holding unit (buffer) that holds the phonetic sound data generated by the speech language processing in the processing unit 103 in the RAM 3. FIG. 4 shows an example of this phonetic data. The sentence “Yesterday was fine weather.” Input in the sentence input unit 101 is the spoken language processing unit 103.
Spoken language processing is performed, and the accent type, the number of mora, the connection information to the next frame, the part of speech, the pause length, the utterance speed, the pitch, and the volume for each of the three frames "Yesterday", "Good" and "It was a weather." It generates phonetic data consisting of various parameters such as.

Reference numeral 105 is a prosody control unit for generating prosody data such as a pitch pattern by referring to the phonetic data stored in the phonetic data storage unit 103, and 106 is the prosody data obtained by the prosody control unit 105. It is a prosody data holding unit (buffer) held in the RAM 3.

A prosody data display unit 107 displays the prosody data held in the prosody data holding unit 105 on the screen of the display unit 7. Reference numeral 108 denotes a prosody data correction unit for correcting and inputting prosody data such as a pitch pattern displayed on the prosody data display unit 107, and reference numeral 109 denotes a prosody data correction value for holding the value corrected and input from the prosody data correction unit 108 in the RAM 3. It is a holding unit (buffer). Reference numeral 110 denotes a correction information label input unit for inputting the label of the value held in the prosody data correction value holding unit 109, and 111 denotes the correction value label input from the correction information label input unit in the RAM 3
It is a correction information label holding unit (buffer) held in.
A correction information holding execution unit 113 holds the value held in the prosody data correction value holding unit 109 and the label held in the correction information label holding unit 111 in the correction information holding unit 112 of the external memory 5 in association with each other. . The correction information holding execution unit is activated by inputting an execution instruction from the input unit 6.

The prosody data stored in the prosody data storage unit 105 is displayed on the display screen of the display unit 7 by the prosody data display unit 107 as shown in FIG. This display is displayed so that the relationship between the phonetic notation "Kinoha Yoi Tenkidsita" and the prosody can be read, and the prosodic modification points are indicated by "○" to make it easier to perform prosody modification operations. To do.

To modify the prosody, while watching this display screen,
Enter the correction value by dragging ○ with the mouse to move it. For example, when you want to change the standard type "Yesterday was a nice day." So, drag the ○ corresponding to “Ha” of “Kinoha” to the top with the mouse, and raise the ○ corresponding to “A” of “Yoi” and the ○ corresponding to “Tenkiddesita” in the same way. . Furthermore, in order to bring it closer to the Kansai style, we also modified the prosody of other parts,
The intonation shown in FIG. 6 is completed.

In order to complete the desired intonation, the execution of vocalization is instructed from the input unit 6 even during the prosody modification work, so that the vocalization in the prosody state displayed on the display screen at that time is executed. The operator hears this utterance and proceeds with the correction work. Since the correction value input by dragging the mouse with a circle is held in the prosody data correction value holding unit 109 at any time, the prosody data held in the prosody data correction value holding unit 109 in response to the input of the utterance execution instruction. By generating the voice data with reference to, and uttering the voice data, it is possible to utter with the intonation corresponding to the prosody data on the display screen.

When the correction of the prosody data is completed, "Kansai style" is input as label information to the correction data from the correction information label input section 110, and the input correction information label is input to the correction information holding section together with the prosody data. Retained at 112. The correction information label functions as an identifier when reading the prosody data from the correction information holding unit 112,
As shown in FIG. 6, when the prosody data is displayed on the display unit 7, the correction information label is displayed together with the prosody data so that the correction data and what label it holds are displayed.

Reference numeral 114 is a usage information label input section for inputting the label of the correction information to be used, and 115 is the label of the usage information input from the usage information label input section 114.
It is a usage information label holding unit (buffer) held in AM3. Reference numeral 116 reads the correction information corresponding to the label of the usage information held in the usage information label holding unit 115 from the correction information holding unit 112, and the prosody data correction value holding unit 109.
It is a usage information reading execution unit held in. The usage information reading execution unit is activated by inputting an execution instruction from the input unit 6.

Reference numeral 117 denotes a prosody data correction value holding unit 109.
A voice generation unit that performs voice generation processing such as prosody, parameter connection, and waveform connection by referring to the value held in
Reference numeral 18 denotes a voice parameter holding unit (buffer) that holds the parameters of the synthetic voice generated by the voice generating unit 117 in the RAM 3. A DA converter 119 converts the voice parameter held in the voice parameter holding unit 118 into synthetic voice. Also, 120 controls the start of execution of voice generation in the voice generation unit 117.

Next, the operation of the present apparatus will be described. FIG. 3 is a flow chart showing the operation procedure of the speech synthesizer of this embodiment.

In FIG. 3, steps S201 to S20.
7 is a state of waiting for input. The state of waiting for input is repeated until one of sentence input, correction of prosody data, correction information label input, correction information retention execution, use information label input, use information read execution, and utterance execution is input. Start each process.

When a sentence for speech synthesis is input to the sentence input unit 101, the input sentence is held in the input sentence holding unit 102, and the process proceeds from step S201 to step S208. In step S208, the phonetic language processing unit 103 performs phonetic language processing on the input sentence held in the input sentence holding unit 102, and the phonetic data as a result is phonetic data holding unit 1.
04. In step S209, the prosody control unit 105 refers to the phonetic data stored in the phonetic data storage unit 104 to generate prosody data such as a pitch pattern, and stores the prosody control information in the prosody data storage unit 109. , And move to step S210. In step S210, the value of the prosody data such as the pitch pattern held in the prosody data holding unit 109 is displayed on the prosody data display unit 107. Then, move to step S201,
Move to the input waiting state.

When the modification value of the prosody data is input from the prosody data modification unit 108, the process proceeds from step S202 to step S211. If the value is held in the prosody data holding unit 106 in step S211, the process proceeds to step S212, and the correction value input from the prosody data correction unit 108 is held in the prosody data correction value holding unit 109. On the other hand, if the prosody data is not stored in step S211, the process proceeds to step S213, and, for example, an error message “Sentence has not been input. Please input.” Is output (display or voice output).
Then, the process returns to step S201.

When the label is input from the correction information label input unit 110, steps S203 to S214 are performed.
Move on to In step S214, the label of the correction information input from the correction information label input unit 110 is held in the correction information label holding unit 111, and the process returns to step S201.

When the instruction to hold and execute the correction information is input from the prosody information holding / executing unit 113, the process proceeds from step S204 to step S215. In step S215, it is checked whether or not the prosody data correction value holding unit 109 and the correction information label holding unit 111 hold the correction value and the label, respectively. If the correction value and the label are held, the process proceeds to step S216, and the correction value held in the prosody data correction value holding unit 109 is set to the correction information label holding unit 11
The label is held in the correction information holding unit 112 together with the label held in 1. Then, the process returns to step S201. If at least one of the correction value and the label is not held in step S215, the process proceeds to step S217 and, for example, "the correction value or the label is not input.
Please enter. Error message,
Then, the process proceeds to step S201.

When the label of the information to be used is input from the use information label input unit 114, the process proceeds from step S205 to step S218. In step S218, the label input from the usage information label input unit 114 is held in the usage information label holding unit 115, and the process proceeds to step S201.

When an instruction to read and use a pattern to be used is input to the usage information read execution unit 116, the process proceeds from step S206 to step S219. Step S
In 219, it is determined whether or not the label used in the usage information label holding unit 115 is held, and if held, the process proceeds to step S220. Then, in step S 220, the correction information corresponding to the label held in the usage information label holding unit 115 is read from the correction information holding unit 112 and held in the prosody data correction value holding unit 109. Then, it returns to step S201. If the label to be used in step S219 is not held, the process proceeds to step S221, for example, an error message "The label has not been entered. Please enter it." Is output, and the process proceeds to step S201. .

When an instruction to execute the utterance is input from the utterance executing unit 120, the process proceeds from step S207 to step S222. In step S222, it is determined whether or not the prosody data holding unit 107 holds the prosody data. If the prosody data is held, the process proceeds to step S224.

In step S224, the voice generator 117
In the above, with reference to the prosody data held in the prosody data holding unit 107 and the correction value held in the prosody data correction unit 109, voice generation processing such as prosody, parameter connection, waveform connection, etc. is performed, and the result is obtained. The voice parameter is held in the voice parameter holding unit 118. Finally, in step S225, the DA converter 119 converts the voice parameter held in the voice parameter holding unit 118 into synthetic voice and outputs the synthetic voice, and then in step S2.
Return to 01.

On the other hand, if the prosody data storage unit 106 does not store the prosody data in step S222, the process proceeds to step S223, and an error message "Sentence has not been entered. Please enter" is displayed. Output and return to step S201. Incidentally, step S
The error messages in 213, S217, S221, and S223 are displayed as characters on the display unit 7 or output as voice messages from the voice output unit 4.

According to the control procedure as described above, desired correction can be added to the prosody data obtained by performing the voice synthesis process and the prosody control process on the input sentence. Therefore, the prosody data such as the pitch pattern can be directly changed. An example of modifying the prosody data is shown below.

FIG. 6 shows an example of a pitch pattern in which the standard pitch pattern for the input sentence "Yesterday was fine weather" shown in FIG. 5 is corrected to the Kansai dialect. Further, FIG. 7 shows an example of a standard (question-like) pitch pattern for the input sentence “is it so”?
Shows an example of a pitch pattern which is modified to a convincing tone, a suspicious tone, and a hammering tone, and is given the label.

As described above, by editing the prosody data that is the result of the prosody control unit and using the edited values to generate speech, various utterances can be generated even for input sentences of the same character. Can be realized. Further, the edited value can be reused by giving a label with the edited value as a pattern and holding / reading.

In the above description, an example in which the prosody data is always corrected has been described, but it is also possible to be able to utter a synthetic voice without correction. In that case, a method of directly designating and using the value held in the prosody data holding unit 106, or holding the value in the prosody data holding unit 106, and at the same time holding the value in the prosody data correction value holding unit 109, a method for uttering synthetic speech using this value, or a prosody data correction value holding unit 1
After determining whether or not a value exists in 09, if the value does not exist, the voice generation unit 117 determines that the prosody data holding unit 1
A method of uttering a synthetic voice using the value held in 06 may be used.

In the above description, the prosody data correction value holding unit 109 holds all the prosody data for the corrected sentence, but only the difference from the value held in the prosody data holding unit 106 is stored. You may keep it.

Further, in the above description, only one set of modified values of prosody data is held in the buffer at the same time, but a set of a plurality of values is held in the buffer and the set of values from the input unit 6 is input. The correction value used for speech synthesis may be switched according to the switching instruction.

In the above description, the input sentence holding unit 10
Although it is described that the speech language processing is performed immediately after the sentence is held in 2, the instruction to execute the speech language processing is given by the input unit 6, and the speech language processing is started according to the instruction. May be.

In the above description, the prosody control process is performed immediately after the sentence is held in the phonetic data holding unit 104. However, the input unit 6 gives an instruction to execute the prosody control process. Alternatively, the prosody control process may be started in response to the instruction.

Further, in the above description, all of the input is manually performed, but other outputs such as a natural language generation device or a natural language dialogue device may be used as the input as they are.

In the above description, the sentence for correcting the prosody is input from the sentence input unit 101, but the input of the sentence is not limited to this, and the sentence is displayed on the normal sentence input screen. A desired sentence or character string in the written sentence may be specified by the keyboard or mouse, and the specified sentence or character string may be captured by the sentence input unit 101.

Further, not only the standard type intonation but also a plurality of types of intonations can be created for a sentence of the same character, and a label can be given and held, so that a speech language process from the input of a sentence to the utterance of that sentence is performed. At the stage of
For example, a sentence with “?” At the end of the sentence is determined to be an interrogative form, and prosodic data to which a label that can identify the intonation of the interrogative form is attached is read from the correction information holding unit 112. Each time you utter, you can utter automatically with a question intonation without modifying the intonation.

[0061]

As described above, according to the present invention,
Displaying the prosody data of the input character string, modifying the displayed prosody data, and voicing the character string according to the modified prosody data, whereby each operator sets and outputs a desired prosody. You can

As described above, according to the present invention, it is possible to correspond to any character string desired by the operator by subjecting the input character string to speech language processing to generate the prosody data to be displayed. You can

As described above, according to the present invention, when displaying the prosody data, by displaying the phonetic notation corresponding to the prosody data in association with the prosody data, which prosody data can be corrected. It is possible to inform the operator whether or not to do so, and the correction operation becomes easy to understand.

As described above, according to the present invention, the correction of the prosody data is performed by moving the mark representing the prosody data on the display screen, so that a beginner can input the correction value of the prosody data. Can be done easily.

As described above, according to the present invention, the character string is the character string specified in the displayed text, so that the work of another routine for inputting the correction data is restarted. Without doing, it is possible to perform the correction work even for the sentence to be processed.

As described above, according to the present invention, the modified prosody data can be reused by adding a label to the modified prosody data and registering it.

As described above, according to the present invention, by inputting a label, reading the prosody data corresponding to the input label, and uttering according to the read prosody data, desired prosody data is obtained. Can be read quickly.

As described above, according to the present invention, the corrected prosody data is held, and the utterance according to the held prosody data is executed according to the instruction of utterance, so that the correction process is performed. You can check your utterance.

As described above, according to the present invention, by controlling the display means for displaying the prosody data, the prosody data can be confirmed by the display means.

As described above, according to the present invention, the utterance of the character string can be confirmed by controlling the utterance of the character string by the speaker.

As described above, according to the present invention, the operator can input the correction data by himself / herself by processing the correction data of the prosody data input from the input means.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of a speech synthesizer.

FIG. 2 is a block diagram showing a functional configuration of a speech synthesizer.

FIG. 3 is a flowchart showing a voice synthesis processing operation procedure.

FIG. 4 is a diagram showing an example of phonetic data for an input sentence “Yesterday was fine weather”.

FIG. 5 is a diagram showing an example of prosody data for an input sentence “Yesterday was fine weather”.

FIG. 6 is a diagram showing a modification example of the input sentence of FIG. 5 in which prosody data is made into a Kansai dialect.

[FIG. 7] A diagram showing an example of prosody data for an input sentence “is it?”

FIG. 8 is a diagram showing a modification example in which prosody data is convincingly applied to the input sentence of FIG. 7.

FIG. 9 is a diagram showing a modification example in which the prosody data is suspicious for the input sentence of FIG.

FIG. 10 is a diagram showing a modification example in which prosodic data is adjusted to the input sentence of FIG.

Claims (20)

[Claims] 1. A voice information processing method comprising: displaying prosody data of an input character string, modifying the displayed prosody data, and uttering a character string in accordance with the modified prosody data. 2. The voice information processing method according to claim 1, wherein the input character string is subjected to speech language processing to generate the prosody data to be displayed. 3. The voice information processing method according to claim 1, wherein when the prosody data is displayed, a phonetic notation corresponding to the prosody data is displayed in association with the prosody data. 4. The voice information processing method according to claim 1, wherein the modification of the prosody data is performed by moving a mark representing the prosody data on a display screen. 5. The voice information processing method according to claim 1, wherein the character string is a character string specified in a displayed sentence. 6. The voice information processing method according to claim 1, wherein a label is added to the modified prosody data and registered. 7. The voice information processing method according to claim 1, wherein a label is input, prosody data corresponding to the input label is read, and a voice is uttered according to the read prosody data. 8. The voice information processing method according to claim 1, wherein the modified prosody data is held, and utterance is executed in accordance with the held prosody data in response to an utterance instruction. 9. The voice information processing method according to claim 1, further comprising controlling display means for displaying the prosody data. 10. The voice information processing method according to claim 1, wherein control is performed so that the speaker utters the character string. 11. The voice information processing method according to claim 1, wherein modified data of the prosody data input from the input means is processed. 12. Input means for inputting a character string, display means for displaying prosody data of the input character string, prosody modification means for modifying the displayed prosody data, and the modified prosody data. And a voicing control means for controlling to utter a character string according to the above. 13. The voice information processing apparatus according to claim 12, further comprising voice language processing means for performing voice language processing on the input character string to generate the prosody data to be displayed. 14. The voice information processing apparatus according to claim 12, wherein the display unit displays the phonetic notation corresponding to the prosody data in association with the prosody data. 15. The voice information processing apparatus according to claim 12, wherein the prosody modification unit performs the prosody modification according to movement of a mark representing the prosody data designated on the display screen. 16. The voice information processing apparatus according to claim 12, wherein the input unit inputs a character string specified in a displayed sentence. 17. The voice information according to claim 12, further comprising registration means for inputting a label by the input means, and adding the input label to the corrected prosody data to register the same. Processing equipment. 18. The voice information according to claim 12, wherein the reading means reads the prosody data corresponding to the label input by the input means, and the voicing control means utters according to the read prosody data. Processing equipment. 19. A holding means for holding the modified prosody data, a voicing instruction means for instructing utterance, and the utterance control means for the prosody held by the holding means in response to an instruction from the voicing instruction means. 13. The voice information processing apparatus according to claim 12, wherein the voice information processing apparatus executes utterance according to the data. 20. The voice information processing apparatus according to claim 12, further comprising a speaker that outputs a voice generated by control of the utterance control means.