JPH0594196A - Speech synthesizing device - Google Patents

Abstract

(57) [Summary] [Object] The present invention is intended to realize a speech synthesizer capable of synthesizing a synthetic speech with less deterioration in quality compared to an actual human speech and without discomfort. [Structure] A complex cepstrum analysis, which is a high-quality Pitch transform method, is used in the analysis and synthesis of real speech, and impulses and vocal tracts are used as sound source information in the voiced part of each voice unit data used in the synthesis of speech. By having multiple sets of required unit response waveforms as characteristics, and having a cutout waveform of the actual voice in the unvoiced part, it is close to human voice without causing distortion of the spectrum envelope due to changes in pitch pattern. It is possible to arbitrarily generate high-quality synthetic speech.

Description

Detailed Description of the Invention

[0001]

[Table of Contents] The present invention will be described in the following order. Field of Industrial Application Conventional Technology Problem to be Solved by the Invention Means for Solving the Problem (FIGS. 1 and 2) Action (FIGS. 1 and 2) Example (1) Principle of Example (2) Implementation Example speech synthesizer (FIGS. 1 and 2) (3) Other embodiments Effect of the invention

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech synthesizer, and is particularly suitable for application to a speech synthesizer using a rule synthesis method.

[0003]

2. Description of the Related Art Conventionally, in a speech synthesizing apparatus based on a rule synthesizing method, after analyzing a series of input characters and synthesizing parameters according to a predetermined rule, it is possible to synthesize speech with any words. (Japanese Patent Laid-Open No. 3-119396).

That is, the speech synthesizer based on the rule synthesizing method analyzes an input character sequence, detects an accent for each phrase according to a predetermined rule, and detects the accent of each phrase as a whole character sequence. Pitch parameters that express intonation, porcelain, etc. are synthesized.

Further, the voice synthesizer similarly divides each clause into voice units such as CV units according to a predetermined rule, and then generates a synthesis parameter expressing the spectrum. As a result, a synthesized sound is produced based on the pitch parameter and the synthesis parameter.

[0006]

By the way, in a conventional residual-based synthesis method using linear prediction analysis as a general synthesis method based on synthesis parameters, when the pitch of synthesized speech is changed, the prediction of the sound source information is performed. The residual waveform is processed.

That is, when the pitch frequency of the synthesized speech is lowered, a constant value 0 is inserted from the end of the prediction residual waveform so that the desired pitch period is obtained, and when the pitch frequency of the synthesized speech is raised. It is a process of setting the desired pitch cycle by cutting off the prediction residual waveform on the way.

However, in the linear predictive analysis, since the source information (pitch information) and the vocal tract characteristic (spectral envelope) are not completely separated from the actual voice, the waveform is included in the predictive residual waveform by this waveform processing. This results in distortion of the existing spectrum envelope information, and this distortion tends to deteriorate the quality of synthesized speech.

The present invention has been made in consideration of the above points, and it is an object of the present invention to propose a voice synthesizing device capable of uttering a synthetic voice with less deterioration in quality than an actual human voice and having no discomfort. To do.

[0010]

In order to solve such a problem, in the first invention, analysis information storage for accumulating a word, a bunsetsu boundary, and a basic accent obtained by analyzing a series of input characters With respect to the unit 3 and the voiced portion having the periodicity in the voice unit, the voice waveform data corresponding to each one-pitch cycle obtained by the analysis process of the real voice is accumulated as the voice unit, and the voice waveform data is stored in the voice unit. As for the unvoiced portion having no periodicity, the memory unit 2 that stores the actual voice as it is as voice waveform data and the analysis information storage unit 3
A voice synthesis rule unit 4 for generating a pitch pattern in accordance with a predetermined phonological rule and a prosodic rule based on the analysis information of 1. and a voice unit of the memory unit 2 and a voice synthesis based on the pitch pattern of the voice synthesis rule unit 4. The voice synthesizing unit 5 is provided.

Further, in the second aspect of the invention, the sentence analysis unit 3 is provided for analyzing the series of characters of the sentence input to the analysis information storage unit 3 into words, bunsetsu boundaries and basic accents.

[0012]

The complex cepstrum analysis, which is a high-quality Pitch transform method, is used in the analysis and synthesis of the real voice, and the impulse and vocal tract as the sound source information in the voiced part of each voice unit data used for the voice synthesis. By having the required multiple sets of unit response waveforms as characteristics, and having the cutout waveform of the actual voice in the unvoiced part, it is possible to obtain a high-level sound close to that of human voice without causing distortion of the spectrum envelope due to the change of pitch pattern. It is possible to realize a voice synthesizing device that can arbitrarily generate high quality synthetic voice.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail with reference to the drawings.

(1) Principle of the embodiment In the case of this embodiment, the complex cepstrum analysis is used to separate the sound source information and the vocal tract characteristics in the voiced part of the real speech in the analysis processing for each voice unit used for synthesis. Is extracted as an impulse, and the vocal tract characteristics are extracted as a unit response of impulse which is sound source information.

This complex cepstrum analysis is an analysis method known as a high-quality pitch conversion method and a speech rate conversion method in the analysis and synthesis of real speech, and the analysis method useful in this analysis and synthesis of speech is a rule of arbitrary sentence utterance. It is designed to be used for synthesis.

The voiced part of the voice unit used for synthesis is
Both the impulse and the unit response extracted by the complex cepstrum analysis method are set as one combination, and as many combinations as the number of frames required for the voice unit voiced portion are stored as voiced portion data. Further, in the unvoiced portion of each voice unit, the unvoiced portion of the actual voice is cut out as it is and stored as data.

As a result, the voice unit is composed of a combination of a plurality of frames consisting of an impulse and its unit response, a cutout waveform of a real voice which is an unvoiced portion, or both.

Therefore, first, together with the voice unit storage unit in which voice unit data having such contents is stored in the memory, the sequence of input characters is analyzed to detect the boundaries of words and phrases and basic accents. A sentence analysis unit is provided.

In addition to this, based on the detection result of the sentence analysis unit, a pitch pattern of synthetic speech is generated according to a predetermined prosody rule, and synthetic waveform data necessary for synthetic speech is also generated according to the phonological rule. Is provided from the voice unit storage unit, and a voice synthesis unit that generates a synthetic voice based on the synthetic waveform data and the pitch pattern.

In this way, the impulses in the voice unit data are sequentially arranged at intervals of the pitch cycle corresponding to the desired pitch pattern, and the unit responses in combination with the respective impulses are superposed one by one. However, even if the pitch period expands or contracts, it has almost no effect on the sound source information because the sound source information is an impulse, and even if the pitch pattern changes significantly, the spectrum envelope is distorted. Does not occur.

As described above, by using the complex cepstrum analysis suitable for the pitch conversion of speech for the rule synthesis, a high quality arbitrary synthesized speech close to human speech can be obtained. Further, since it does not require a complicated calculation process unlike the synthesizing method using the synthesizing parameter, it is possible to speed up the process in the voice synthesizing unit.

(2) Speech synthesizer of the embodiment In FIG. 1, reference numeral 1 shows a schematic configuration of a speech synthesizer having an arithmetic processing unit as a whole, a voice unit storage unit 2, a sentence analysis unit 3, and a speech synthesis rule unit 4. And the speech synthesis unit 5, and the sentence analysis unit 3 first inputs a text input from a predetermined input device (consists of a sentence represented by a sequence of characters).
Is analyzed based on a predetermined dictionary, converted into a kana character string, and then decomposed into words and phrases.

In other words, in Japanese, words are not separated into words like English, so words such as "US industry" are "US / industry / world" and "US /
There are two types, such as "domestic / industry".

For this reason, the sentence analysis unit 3 is designed to decompose the text input into words and phrases by using the continuity of words and the statistical property of words while referring to the dictionary. Detect the boundaries of. Further, the sentence analysis unit 3 detects basic accents for each word and then outputs them to the speech synthesis rule unit 4.

The voice synthesis rule unit 4 is adapted to process the detection result of the sentence analysis unit 3 and the text input according to a predetermined phonological rule set based on the characteristics of Japanese. That is, it is known that natural Japanese speech can be divided into about 100 voicing units when distinguished based on linguistic characteristics. For example, when the word "Sakura" is divided into voicing units. , “Sa” + “ku” + “ra” can be divided into three CV units.

Further, in Japanese, when words are continuous, the initial syllable of the succeeding words becomes dull (that is, it becomes continuous), and the ga-sound other than the beginning becomes nasal. There is a feature that the utterance changes with the case.

Accordingly, the speech synthesis rule unit 4 is configured to set the phonological rules according to the characteristics of Japanese, and the text input according to the phonological rules is a phonological symbol string (that is, "sa" + "mentioned above). (Consecutive columns such as "ku" + "ra")
It is designed to be converted into. Further, the voice synthesis rule unit 4 loads data of each voice unit from the voice unit storage unit 2 based on the phoneme symbol string.

Here, in the voice synthesizer 1, a synthesized voice is produced by using a waveform editing method.
The data loaded from the voice unit storage unit 2 is waveform data used when generating a synthetic sound represented by each CV unit. The voice unit data used for this waveform synthesis has the following configuration.

With respect to the voiced part of the voice unit data, the impulse corresponding to one pitch and the unit response waveform extracted by using the complex cepstrum analysis in the voiced part of the real voice are set as one set, and this set is set as one voice. The unvoiced part of the voice unit data is cut out from the waveform of the unvoiced part of the actual voice and stored as it is.

Therefore, when the voice unit data is in CV units, when the consonant part C of one voice unit CV is an unvoiced consonant, the unvoiced part cutout waveform and a plurality of sets of impulses and unit response waveforms are used. One voice unit data is formed, and when the consonant part C of one voice unit CV is a voiced consonant, one voice unit data is formed by only a plurality of pairs of impulses and unit response waveforms. Become.

The voice synthesis rule unit 4 synthesizes the voice unit data loaded from the voice unit storage unit 2 in the order corresponding to the text input (hereinafter, this data will be referred to as synthesized waveform data), and thus, without inflection. , It is possible to obtain a synthetic speech waveform that reads the text input.

Further, the voice synthesis rule unit 4 divides the text input into appropriate lengths based on a predetermined prosody rule to detect a break (that is, a pause). In this way, as shown in FIG. 2, for example, when the text “A beautiful flower was received from Mr. Yamada” is input as the text input (FIG. 2 (A)), the text input is “clean”. , "Hana", "From Yamada-san", and "I got it", then a pose is detected between "Hana" and "From Yamada-san" (Fig. 2 (B)).

Further, the speech synthesis rule unit 4 detects the accent of each phrase based on the prosody rule and the basic accent of each word. That is, the accent of a Japanese phrase alone can be expressed sensuously in two levels, high and low, in units of kana characters (hereinafter referred to as mora). At this time, the accent position of the phrase can be distinguished according to the content of the phrase.

For example, edges, chopsticks, and bridges are 2-mora words, and are classified into 0 type with no accent, 1 type with accent position in the first mora, and 2 type with accent position in 2nd mora. You can Thus, in this embodiment, the speech synthesis rule unit 4 classifies each phrase of the text input into 1 type, 2 type, 0 type and 4 type (FIG. 2 (C)). To do.

Furthermore, the voice synthesis rule unit 4 generates a basic pitch pattern representing the intonation of the entire text input, based on the accent and pause detection results. That is, in Japanese, the accent of a bunsetsu can be sensuously expressed in two levels, while the actual intonation is characterized by gradually decreasing from the position of the accent (FIG. 2 (D)).

Further, in Japanese, when the bunsetsu becomes one sentence continuously, the intonation gradually decreases from one pose to the next (FIG. 2 (E)).
Therefore, the speech synthesis rule unit 4 generates a parameter representing the intonation of the entire text input for each mora based on the characteristics of the Japanese language, and then the intonation changes smoothly as if a human uttered. Then, parameters are set by interpolation between mora.

Thus, the speech synthesis rule unit 4 synthesizes the parameters of each mora and the interpolated parameters in the order according to the text input (hereinafter referred to as a pitch pattern), and thus the pitch representing the intonation of the speech read out from the text input. A pattern (FIG. 2 (F)) can be obtained.

Next, the voice synthesizing unit 5 performs a waveform synthesizing process based on the synthetic waveform data and the pitch pattern to generate a synthetic sound. This waveform synthesis processing is performed as follows. In the voiced part of the synthetic speech, impulses in the synthetic waveform data are arranged based on a pitch pattern, and a unit response waveform corresponding to each of the arranged impulses is superimposed on each impulse.

In the unvoiced part of the synthetic voice, the cut-out waveform in the synthetic waveform data is used as it is as the waveform of the desired synthetic voice. As a result, it is possible to obtain a synthetic sound in which the intonation changes according to the change in the pitch pattern.

Therefore, since the impulse is used for the sound source information in the synthesized sound, even if the pitch period of the synthesized sound expands or contracts, there is almost no effect on the sound source information, and even if the pitch pattern greatly changes, the spectrum is changed. It is possible to obtain a high-quality arbitrary synthesized voice that is similar to human voice without causing distortion in the envelope.

In the above configuration, the text input input from the predetermined input device is analyzed by the sentence analysis unit 2 with reference to the predetermined dictionary, and the word, the boundary of the phrase and the basic accent are detected. The results of detecting the words, the boundaries of the clauses, and the basic accents are processed by the speech synthesis rule unit 4 in accordance with a predetermined phonological rule to generate synthetic waveform data representing a speech in which the text input is read aloud without intonation. It

Further, the detection result of the word and phrase boundaries and the basic accent is processed by the voice synthesis rule section 4 in accordance with a predetermined prosody rule to generate a pitch pattern representing the intonation of the entire text input. The pitch pattern is output to the voice synthesizing unit 5 together with the synthetic waveform data, and a synthetic sound is generated based on the pitch pattern and the synthetic waveform data.

According to the above configuration, in each voice unit data used for synthesizing a voice, the voiced portion has a plurality of required sets of impulses as sound source information and unit response waveforms as vocal tract characteristics. Since the unvoiced part has the cut-out waveform of the actual voice, a voice synthesizer capable of arbitrarily generating high-quality synthesized voice close to human voice without causing distortion of the spectrum envelope due to change in pitch pattern is provided. realizable.

(3) Other Embodiments In the above-mentioned embodiment, the waveform is synthesized by superimposing the unit response waveform by using the impulse as the sound source information, but the impulse is ideal. By considering it as an impulse, a desired synthesized sound may be generated by arranging the unit response waveforms directly so as to correspond to the pitch pattern without arranging and superimposing the impulses at pitch period intervals.

Further, in the above-described embodiment, the voice unit data is held in the voice unit storage unit 2 in the unit of CV. However, the data is held not only in the unit of CV but also in another unit of voice such as CVC unit. May be.

[0046]

As described above, according to the present invention, a complex cepstrum analysis, which is a high-quality Pitch transform method, is used in the analysis and synthesis of real voice, so that a high-quality synthesized voice close to human voice is obtained. It is possible to realize a voice synthesizing device capable of synthesizing any voice.

[Brief description of drawings]

FIG. 1 is a block diagram showing a speech synthesizer according to an embodiment of the present invention.

FIG. 2 is a schematic diagram for explaining the operation.

[Explanation of symbols]

1 ... Voice synthesis device, 2 ... Voice unit storage unit, 3 ... Text analysis unit, 4 ... Voice synthesis rule unit, 5 ... Voice synthesis unit.

Claims (2)

[Claims] 1. An analysis information storage unit for storing a word, a segment boundary, and a basic accent obtained by analyzing a sequence of input characters, and a voiced portion having periodicity in a voice unit, a real voice. The voice waveform data corresponding to each one-pitch cycle obtained by the analysis processing of step 1 is stored as the voice unit, and for the unvoiced portion having no periodicity in the voice unit, the actual voice is directly used as the voice waveform. A memory unit for storing as data, a voice synthesis rule unit for generating a pitch pattern according to a predetermined phonological rule and a prosodic rule based on the analysis information of the analysis information storage unit, the voice unit and the voice of the memory means. A voice synthesizing device comprising: a voice synthesizing unit for synthesizing a voice based on the pitch pattern of the synthesis rule unit. 2. The speech analysis system according to claim 1, wherein the analysis information storage unit includes a sentence analysis unit that analyzes a sequence of characters of an input sentence into words, boundaries between phrases, and basic accents. Synthesizer.