JPH01211798A - Regular synthesizing device for voice - Google Patents

Abstract

PURPOSE:To add individual features of a 2nd speaker while minimizing individual voice information which is prepared by providing a regular synthesizing means which synthesizes the voice signal of a 1st standard speaker and a voice converting means which converts the voice signal of the 1st speaker into the voice signal of the 2nd speaker to whom the individual features are to be added. CONSTITUTION:The regular synthesis part 20 including a unit set information file 21 of voice receives a character information signal s1 and refers to the unit set information file 21 of voice to synthesize the voice signal s2 of the standard speaker A. Then the voice quantity conversion part 30 receives the voice signal s2 of the speaker A and refers to a previously registered voice individual information file 40 of the speaker B whose voice is to be given individuality to convert the voice quality of the speech signal (regularly synthetic voice) of the speaker A into the voice quality of the speaker B, thereby outputting the voice signal s4 of the speaker B. Consequently, while the individual voice information is minimized, the individuality of the speaker is given to the regularly synthetic voice.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a rule-based synthesis device for speech, and particularly to a rule-based synthesis device that converts and outputs sound quality.

[Prior Art and Problems to be Solved by the Invention] Conventional rule synthesis systems output synthesized sounds of one or several types (male voice, female voice, childlike voice, and elderly voice). etc.) are limited to However, when a rule synthesis system is actually used, it is desirable that the sound quality of the output synthesized voice also have individual characteristics, just as each person has individual characteristics in their voice.

Apart from this, there is also a demand for the rule-synthesized speech to have the desired sound quality of a specific person.

However, in conventional rule synthesis systems, (1) the speech units used for rule synthesis are created from hundreds of voices uttered by a given speaker, which places a heavy burden on the speaker;

(2) Creation of speech units is difficult and labor-intensive to perform fully automatically, and it is virtually impossible to create speech units for many speakers.

(3) If a unit set of speech used for rule synthesis is created for each speaker, the amount of memory to be stored will become enormous.

For the following reasons, there was a problem in that the above requirements could not be realized.

This invention was made in order to solve the above-mentioned problems, and it is possible to reduce the burden on the speaker, the data creator for rule synthesis, and the rule synthesis system as much as possible, while adding the burden on the speaker to the rule synthesized speech. The purpose is to give it individuality.

[Means for Solving the Problems] A speech rule synthesis device according to the present invention receives a character information signal from the outside, and by referring to a standard set of speech unit signals of a first speaker, synthesizes a first speech unit signal. A rule synthesis means for synthesizing speech signals of speakers, and a conversion device that stores conversion signals necessary for converting the speech of a first speaker into the speech of a second speaker who wants to add personal characteristics to the speech. It includes a signal file means, and a voice conversion means for converting the synthesized first speaker's voice signal into a second speaker's voice signal based on the converted signal stored in the converted signal file means.

[Operation] The speech rule synthesis device according to the present invention performs rule synthesis of the standard first speaker's speech signal by the rule synthesis means, and then synthesizes the voice signal by the voice conversion means according to the converted signal stored in the converted signal file means. , the first speech signal is converted into the speech signal of the second speaker having individual characteristics, so there is no need to prepare a set of speech units necessary for regular synthesis for each individual with different speech characteristics. .

[Embodiment of the Invention] FIG. 1 is a block diagram showing an embodiment of a rule synthesis apparatus according to the invention.

Referring to FIG. 1, a character information signal s1 including character strings such as characters and accent types, and a prosody signal is applied to input unit 10 from the outside. A rule synthesis unit 20 including a voice unit set information file 21 generates a character information signal s1.
, and synthesizes a standard speech signal s2 of speaker A with reference to the speech unit set information file 21. Here, the speech unit set information file 21 is a database in which information about the standard speaker A, which is a speech unit such as a phoneme or a syllable, is stored in advance.

The voice quality converting unit 30 receives the voice signal s2 of the speaker A, refers to a pre-registered voice personal information file 40 of the speaker B who wants to add individuality to the voice, and converts the voice signal s2 of the speaker A ( Ruled synthesized speech) Convert the voice quality of s2 to the voice quality of speaker B,
It is output as speaker B's audio signal S4.

As a voice quality conversion method in the voice quality conversion unit 30, a voice quality conversion method using vector quantization is used. This method converts the voice quality between a standard speaker A, which is the standard for the rule synthesis unit 20, and a speaker B, who wants to add individuality to the voice, using a conversion code that is a correspondence between the codebooks of each speaker. This is done using a book.

The conversion codebook includes the power, pitch frequency, and spectrum information of the speaker's voice to which individuality is to be added, and the characteristics of the voice are expressed discretely. The audio personal information file 40 in FIG. 1 includes the contents of this conversion codebook.

FIG. 2 is a flow diagram showing the procedure for creating a conversion codebook.

Referring to Figure 2, below is the conversion code book 41.42
．． The procedure for finding 43 will be explained.

First, in steps 301 and 302, the voices of speaker A and speaker B are subjected to LPG analysis to obtain power, pitch frequency, and spectral parameters. Next, the spectral parameters are vector quantized in steps 30B and 304, the power is scalar quantized in steps 305 and 306, and the power is scalar quantized in steps 30B and 304.
7 and 308, the pitch frequency is scalar quantized.

Step 309 uses the spectral parameters to take the time correspondence of the voices uttered by speaker A and speaker B.
DP matching is performed using the Double 5plit method. Based on the time-related information obtained here, in steps 3101311 and 312,
The correspondence between speaker A and speaker B is determined for each feature, and a histogram is created. The spectral parameter and power conversion codebooks 41 and 43 are obtained by linear combination of feature vectors of speaker B using this hisdogram as weight. Further, the pitch frequency conversion codebook 42 is created using the feature vector of speaker B that gives the maximum value of this histogram.

FIG. 3 is a flow diagram showing the voice quality conversion procedure in the voice quality conversion section 30.

A voice quality conversion method using a conversion codebook will be described below with reference to FIG. The speech signal s2 of speaker A is subjected to LPG analysis in step 401 to extract power, pitch frequency and spectral parameters.

Next, the spectral parameters from speaker A's spectral codebook are vector quantized in step 402, the powers from speaker A's power codebook are scalar quantized in step 403, and step 40
4, the pitch frequencies from speaker A's pitch frequency codebook are scalar quantized. In the process of decoding these quantized parameters, the aforementioned transformation codebooks 41.42.43 are used. That is, in step 405, the spectrum conversion codebook 41 from speaker A to speaker B is used, in step 406, the power conversion codebook 43 is used, and in step 407, the pitch frequency conversion codebook 42 is used. Then, the speech signal s4 of speaker B is synthesized in step 408 using each of the converted parameters.

FIG. 4 is a schematic block diagram showing the hardware configuration of a rule synthesis system including a rule synthesis apparatus according to the present invention.

Referring to FIG. 4, this rule synthesis system consists of amplifier 1
, a low-pass filter 2 , an A/D converter 3 , and a computer system 4 . The amplifier 1 is for amplifying an input audio signal, and the low-pass filter 2 is for removing aliasing noise from the amplified audio signal. A/D
The converter 3 converts the audio signal into a 16-bit digital signal using a 12 kHz sampling signal. The computer system 4 includes a rule synthesis device (arithmetic processing unit) 5, a magnetic disk 6, terminals 7, and a printer 8. The speech rule synthesis device according to the present invention is constructed within the rule synthesis device 5 shown in FIG.

[Effects of the Invention] As described above, according to the present invention, there is provided a rule synthesis means for synthesizing a standard voice signal of the first speaker, and a method for imparting personal characteristics to the voice signal of the first speaker. and a voice converting means for converting the voice signal into the voice signal of the second speaker, so that the synthesized speech of the first speaker, that is, the standard speaker, can be converted into the regular synthesized voice of the first speaker, that is, the standard speaker, while keeping the personal voice information that must be prepared in advance to a minimum. It is possible to add the personal characteristics of the second speaker.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of a rule synthesis device according to the present invention. FIG. 2 is a flow diagram showing the procedure for creating a conversion codebook. FIG. 3 is a flow diagram showing the voice quality conversion procedure in the voice quality conversion section. Figure 4 shows
1 is a schematic block diagram showing the hardware configuration of a rule synthesis system including a rule synthesis device according to the present invention. In the figure, 1 is an amplifier, 2 is a low-pass filter, 3 is an A/D converter, 4 is a computer system, 5 is a rule synthesis device, 10 is an input section, 20 is a rule synthesis section, 21 is an audio unit set information file, 30 is a voice quality conversion unit, 40 is a voice personal information file, 50 is an output unit, sl is a text information signal, s2 is a voice signal of speaker A, s3 is a voice personal information signal of speaker B, and s4 is a voice personal information signal of speaker B. Indicates an audio signal.

Claims (1)

[Scope of Claims] It includes a standard set of voice unit signals of a first speaker, receives a character information signal from the outside, and refers to the set of voice unit signals, thereby generating the voice of the first speaker. a rule synthesis means for synthesizing signals; and a conversion signal file storing conversion signals necessary for converting the voice of the first speaker into the voice of a second speaker whose voice is desired to have personal characteristics. means, and the synthesized first
voice conversion means for receiving the voice signal of the first speaker and converting the voice signal of the first speaker into the voice signal of the second speaker based on the conversion signal stored in the conversion signal file means; A speech rule synthesis device, including: