JP2004126011A - Speech synthesis method and apparatus and speech synthesis program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice synthesizing method which generates natural voiced sounds. <P>SOLUTION: Elementary speech units corresponding to inputted rhythm information are selected from an elementary speech unit dictionary which is stored in an elementary speech unit storage part and consists of a plurality of elementary speech units (S1 to S2), and rhythm control of selected elementary speech units is performed (S3), and phases of voiced sounds included in elementary speech units are randomized (S4), and elementary speech units subjected to rhythm control and phase randomization are connected to generate a synthesized voice signal (S5). <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a speech synthesis method and apparatus used for text synthesis, for example, and more particularly to a voice synthesis technique for voiced sounds.
[0002]
[Prior art]
The rule synthesis method, which is one of the voice synthesis methods, is a technique for automatically generating a synthesized voice signal from input prosody information. In the rule synthesizing method, small units such as syllables, phonemes, and one pitch section stored in the speech unit storage unit (which are referred to as phonemes) are stored in accordance with prosody information such as a phoneme symbol string, a phoneme duration, a pitch pattern, and power. A composite voice signal is generated by selecting the characteristic parameters (referred to as a piece) and connecting them after controlling the pitch and the duration. The speech synthesis technology based on the rule synthesis method is used, for example, for text synthesis that artificially generates speech from an arbitrary sentence (text).
[0003]
In the speech synthesis technology based on such a rule synthesis method, in order to improve the quality of a synthesized speech signal, the characteristic parameters of a representative speech unit include a speech waveform cut out from speech data and a sound source obtained by analyzing the speech waveform. A combination of a signal and a coefficient representing the characteristic of the synthesis filter is often used.
[0004]
For example, in the former speech synthesis technology using a speech waveform cut out from speech data as a speech unit, first, a phoneme symbol string, a phoneme duration, a pitch pattern, and a power obtained by a sentence analysis of a text provided for text synthesis. A speech unit is selected from the speech unit dictionary according to the prosody information such as. After controlling the pitch and the phoneme duration for the selected speech units according to the input prosody information, the speech units are connected to generate a synthesized speech signal.
[0005]
Here, the control of the pitch and the phoneme duration based on the prosody information is realized by repetition or thinning of the unit waveform stored in the speech unit. Generally, a pitch waveform is used as a unit waveform for a voiced sound, and a waveform divided into a predetermined length is used as a unit waveform for an unvoiced sound.
[0006]
[Problems to be solved by the invention]
In the conventional speech synthesis technology, there is a problem that the naturalness of the synthesized speech is reduced when the same unit waveform is repeated. This is because the repetition of the same unit waveform emphasizes the periodicity of the band where the periodicity should be weakened, and as a result, the buzzing feeling of the synthesized voice increases and the naturalness of the synthesized voice is impaired. . Even when the same unit waveform is not repeated, if the adjacent unit waveforms have the same phase (when the periodicity is emphasized) due to a problem in creating a speech unit, the same reason is applied. , The naturalness of the synthesized speech may be reduced.
[0007]
In Japanese Unexamined Patent Publication No. Hei 5-108095 (Patent Document 1), a phase randomization process is performed on a speech unit waveform at the time of generating an unvoiced synthesized speech, and the phase-randomized speech unit waveform is added or superimposed while shifting. Thus, a technique for improving the naturalness of unvoiced sound has been disclosed. However, no effective solution for voiced sounds is known.
[0008]
Accordingly, an object of the present invention is to provide a speech synthesis method and apparatus for generating a natural voiced sound and a speech synthesis program.
[0009]
[Patent Document 1]
JP-A-5-108095
[Means for Solving the Problems]
In order to solve the above-described problem, in the present invention, when a synthesized speech signal is generated by connecting a speech unit selected from a speech unit dictionary including a plurality of speech units, a voiced sound of the synthesized speech signal is generated. A process for randomizing the phase is performed. By randomizing the phase of the voiced sound of the synthesized voice signal, the periodicity of the portion where the periodicity is unnecessarily strong can be weakened, and the synthesized voice signal becomes more natural.
[0011]
The strength of the periodicity is not the same in all bands, and there are some bands where strong periodicity is desirable and some bands where weak periodicity is desirable. If the phase is randomized only in a band where weak periodicity is desired, the synthesized speech signal becomes more natural. In general, if a strong periodicity exists in the high frequency range, a synthesized sound signal often has a buzzing feeling, and an effect can be easily obtained by specifying a band in which phase randomization is performed in the high frequency range. Further, since the strength of the periodicity is not the same for all segments but differs for each segment, further improvement can be obtained by controlling the randomization of the phase for each speech unit.
[0012]
The phase randomization is desirably performed on a unit waveform that has been repeated or thinned out by prosody control, but in order to reduce the amount of calculation when generating a synthesized speech signal, the phase randomization is performed in advance. It is also possible to store the speech unit in the storage unit as a speech unit dictionary, and the periodicity of adjacent unit waveforms of the speech unit may become unnecessarily strong due to problems in speech unit creation. It is effective when it is.
[0013]
A speech synthesis device according to the present invention includes a speech unit storage unit that stores a speech unit dictionary including a plurality of speech units, and a unit that selects a speech unit corresponding to prosody information input from the speech unit dictionary. A segment selection unit, a unit for randomizing the phase of a voiced sound included in the selected speech unit, a prosody control unit for controlling the prosody of the speech unit according to the prosody information, and a speech unit whose prosody is controlled. And a segment connection unit for generating a synthesized voice signal by connection.
[0014]
Further, according to the present invention, a process of selecting a speech unit corresponding to prosody information from a speech unit dictionary composed of a plurality of speech units stored in advance, and a phase of a voiced sound included in the selected speech unit. A speech synthesis program for causing a computer to perform a process of randomizing, a process of controlling the prosody of a speech unit according to prosody information, and a process of connecting a speech unit whose prosody is controlled to generate a synthesized speech signal. Is provided.
[0015]
BEST MODE FOR CARRYING OUT THE INVENTION
(1st Embodiment)
FIG. 1 shows a configuration of a speech synthesis system according to a first embodiment of the present invention. In this speech synthesis system, a speech unit selection unit 11, a speech unit storage unit 12 storing a plurality of speech units constituting a speech unit dictionary, a prosody control unit 10, and a speech unit are connected to perform speech synthesis. It comprises a segment connection unit 14 for generating a signal, a switching unit 17 for switching between voiced and unvoiced sounds, and a phase randomization processing unit 13 for performing a phase randomization process on voiced sounds.
[0016]
The operation of the speech synthesis system according to the present embodiment will be described with reference to a flowchart shown in FIG. First, a sentence analysis of a text to be subjected to text synthesis is performed in a sentence analysis / prosody information generation unit (not shown), thereby generating prosody information such as a phoneme symbol string, a phoneme duration, a pitch pattern, and power. The generated prosody information 100 is input to the prosody control unit 10 and the segment selection unit 11 (step S1).
[0017]
In the unit selection unit 11, the speech unit 104 is selected from the speech unit storage unit 12 according to the input prosody information 100 (step S2). The selected speech unit 104 is supplied to the prosody control unit 10, and the prosody control unit 10 applies prosody control to the speech unit 104 supplied from the unit selection unit 11 according to the input prosody information 100, for example, pitch. And the control of the phoneme duration is performed (step S3).
[0018]
The control of the pitch and the phoneme duration by the prosody control unit 10 is realized by repetition or thinning of the pitch waveform stored in the speech units 301 and 302 in the voiced sound portion as shown in FIG. On the other hand, the control of the phoneme duration in the unvoiced sound portion by the prosody control unit 10 is realized by passing a noise source having the phoneme duration in a synthesis filter having a filter coefficient determined by the speech unit 102.
[0019]
In the speech unit 101 whose pitch and phoneme duration are controlled by the prosody control unit 10 in this manner, the voiced sound portion is subjected to phase randomization by the phase randomization processing unit 13 (step S4).
[0020]
The voice segment 102 of the voiced portion after the phase randomization is performed is supplied to the segment connection unit 14 via the switching unit 17. On the other hand, the unvoiced sound part is supplied from the prosody control part 10 to the segment connection part 14 via the switching part 17. In the segment connection unit 14, the speech unit 102 in which the phase of the voiced sound portion is randomized is connected, and a synthesized speech signal 103 is generated (step S5).
[0021]
Next, the phase randomization processing unit 13 which is a characteristic part of the present embodiment will be described in detail with reference to FIGS.
FIG. 4 shows a specific configuration example of the phase randomization processing unit 13. 4 includes FTT (fast Fourier transform) processors 20 and 24, a noise signal generator 23, a phase converter 21 and an IFFT (inverse fast Fourier transform) processor 22.
[0022]
A unit waveform (hereinafter, referred to as a pitch waveform) 200 corresponding to a voiced sound part of a speech unit is input to the phase randomization processing unit. The pitch waveform 200 is subjected to FFT processing by the FFT processing unit 20, and a Fourier coefficient 201 is generated. On the other hand, the FFT processing unit 24 similarly performs an FFT process on the noise signal 204 generated by the noise signal generation unit 23, and generates a Fourier coefficient 205.
[0023]
The phase converter 21 replaces the phase of the Fourier coefficient 201 corresponding to the pitch waveform 200 with the phase of the Fourier coefficient 205 corresponding to the noise signal 204 in a preset band (a band in which the periodicity is to be weakened). Generates a new Fourier coefficient 202. The Fourier coefficient 202 of the pitch waveform in which the phase of the specific band has been replaced by the phase conversion unit 21 is subjected to IFFT processing by the IFFT processing unit 22 to generate a pitch waveform 203 in which the phase of the specific band is randomized.
[0024]
In the pitch waveform 203 thus obtained, even if the pitch waveform is repeated by the prosody control in the above-described prosody control unit 10, the periodicity is weakened by the effect of the randomization of the phase in the band where the periodicity should be weakened. Therefore, the buzzing feeling of the synthesized voice signal 103 is reduced, and the naturalness is improved.
[0025]
This effect will be described with reference to FIG. FIG. 5A shows an example of a spectrum of a synthesized speech not subjected to the phase randomization process, and FIG. 5B shows an example of a spectrum of a natural speech corresponding thereto. As shown in FIG. 5B, the natural voice has a weak periodicity in the high-frequency portion and is unvoiced. On the other hand, in the synthesized speech signal not subjected to the phase randomization processing, the periodicity in the entire band is strong as shown in FIG. Cause. In the present embodiment, the spectrum of the synthesized voice signal 103 can be made closer to the spectrum of the natural voice by randomizing the phase of the band of the high frequency band. That is, the effect of reducing the buzzer feeling of the synthesized voice signal 103 and improving the naturalness is obtained.
[0026]
FIG. 6 shows another specific configuration example of the phase randomization processing unit 13. Since the strength of the periodicity for each band differs for each speech unit, determining a band for which phase randomization processing is performed for each speech unit can further improve the naturalness of the synthesized speech signal 103. The phase randomization processing unit 13 shown in FIG. 3 is an example in which the phase randomization processing is controlled for each speech unit in this manner, and a band selection unit that selects a band for performing the phase randomization processing in the configuration of FIG. 25 is added.
[0027]
The band selection unit 25 selects a band for which phase randomization processing is to be performed based on information such as phoneme information, and supplies the result to the phase conversion unit 21. It is also possible to store the information of only the band for which the phase randomization processing is performed in the speech unit and directly supply the information to the phase conversion unit 21. In that case, the band selection unit 25 can be omitted.
[0028]
The phase conversion unit 21 replaces the phase of the Fourier coefficient 201 corresponding to the pitch waveform 200 with the phase of the Fourier coefficient 205 corresponding to the noise signal 204 only for the specific band set in this way, and adds a new Fourier coefficient 202 Is generated to perform a phase randomization process.
[0029]
As described above, in the present embodiment, the naturalness of the synthesized voice signal 103 can be further improved by setting the band in which the phase randomization process is performed for each voice unit.
[0030]
(Second embodiment)
FIG. 7 shows a configuration of a speech synthesis system according to the second embodiment of the present invention. The parts corresponding to those in FIG. 1 are denoted by the same reference numerals, and the description will focus on differences from the first embodiment. In the first embodiment, the speech unit stored in the speech unit storage unit 11 is represented by a speech waveform, whereas in the present embodiment, the speech unit has the characteristics of the sound source signal and the synthesis filter. It is represented by a set of coefficients to be represented, and has a configuration in which a synthesis filter processing unit 15 is added after the segment connection unit 14.
[0031]
Therefore, in the present embodiment, the phase randomization processing of the speech unit by the phase randomization processing unit 13 is performed based on the sound source signal before being processed by the synthesis filter processing unit 15, for example, the prosody control output from the prosody control unit 10. This is performed on the speech unit 111 of the residual signal. The speech unit 112 of the prosody-controlled residual signal subjected to the phase randomization process in this way is input to the unit connection unit 14 via the switching unit 17 and the phase randomized residual signal 113 generated here Are processed by the synthesis filter unit 15 to generate a synthesized speech signal 103.
[0032]
As described above, in a speech synthesis system in which a speech unit is represented by a combination of a sound source signal and a coefficient representing a characteristic of a synthesis filter, a phase randomization process of the speech unit is performed on the sound source signal before the synthesis filter processing. It is apparent that the same effects as those of the first embodiment can be obtained.
[0033]
(Third embodiment)
FIG. 8 shows a speech synthesis system according to a third embodiment of the present invention, which is a modification of the second embodiment. The configuration is such that a phase randomization process is performed on the subsequent residual signal 114. It is apparent that the present embodiment also provides the same effects as the first embodiment.
[0034]
(Fourth embodiment)
FIG. 9 shows a configuration of a speech synthesis system according to the fourth embodiment of the present invention. The parts corresponding to those in FIG. 1 are denoted by the same reference numerals, and the description will focus on differences from the first embodiment. In the first embodiment, the phase randomization process is performed on the pitch waveform under the prosody control. In the present embodiment, however, the speech unit 120 in which the phase of the pitch waveform is randomized in advance is subjected to phase randomization. It is configured to be stored in the coded speech unit storage unit 16.
[0035]
According to the present embodiment, by performing the phase randomization process of the speech unit offline in advance, it is not necessary to perform the phase randomization process of the pitch waveform online, so that the calculation amount when generating the synthesized speech signal is reduced. There is an advantage that you can. This embodiment has no effect on the problem that the periodicity of the synthesized speech signal is unnecessarily emphasized due to the repetition of the pitch waveform in the prosody control. It is clear that the effect is obtained when the periodicity of the adjacent pitch waveforms of the speech unit is stronger than necessary, and it is apparent that basically the same effect as in the first embodiment can be obtained.
[0036]
(Fifth embodiment)
FIG. 10 shows a speech synthesis system according to a fifth embodiment of the present invention, which is a modification of the first embodiment, in which a phase randomization processing unit 13 is arranged after a unit connection unit 14 and a synthesized speech signal 105 The phase randomization process is performed on the voiced sound portion of. In this case, the switching unit 17 that switches between voiced sound and unvoiced sound is disposed after the phase randomization processing unit 13. It is needless to say that the same effect as that of the first embodiment can be obtained even when the phase of the voiced sound is randomized for the synthesized voice signal 105 obtained by connecting the voice segments in this way.
[0037]
Although some embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and can be implemented with various modifications. For example, in the above embodiment, the phase randomization processing unit is configured to replace the phase of the Fourier coefficient of the pitch waveform with the phase of the Fourier coefficient of the noise signal. You can also
[0038]
Further, in the above embodiment, the phase randomization processing is performed before the generation of the synthesized voice signal, but the present invention is not limited to this, as long as the configuration is such that the phases of adjacent pitch waveforms can be randomized.
[0039]
Further, in the above embodiment, the configuration is such that the phase randomization process is not performed on the unvoiced sound portion. However, it is sufficient that the configuration is such that the phase randomization is performed on at least the voiced sound portion. May be configured to perform a phase randomization process.
[0040]
The above-described speech synthesis processing based on the present invention can be realized by hardware, but can also be realized by software processing using a computer. Therefore, according to the present invention, it is also possible to provide a program for causing a computer to perform the above-described speech synthesis processing.
[0041]
【The invention's effect】
As described above, according to the present invention, by performing the process of randomizing the phase of the voiced sound of the synthesized voice signal, it is possible to obtain a synthesized voice excellent in naturalness.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration of a speech synthesis system for describing a first embodiment of a speech synthesis method according to the present invention; FIG. 2 is a diagram illustrating a procedure of speech encoding processing in the first embodiment; Flowchart FIG. 3 is a diagram showing a method of controlling the prosody of a speech unit. FIG. 4 is a block diagram showing a specific configuration example of a phase randomization processing unit. FIG. 5 is another specific configuration of the phase randomization processing unit. FIG. 6 is a block diagram showing an example. FIG. 6 is a diagram showing a change in a frequency spectrum in a phase randomization process. FIG. 7 is a block diagram of a unit dictionary encoding system for explaining a second embodiment of the speech synthesis method according to the present invention. FIG. 8 is a block diagram illustrating a configuration of a speech synthesis system for explaining a third embodiment of the speech synthesis method according to the present invention. FIG. 9 is a fourth diagram illustrating a speech synthesis method according to the present invention. An embodiment will be described. Block diagram showing the arrangement of a speech synthesis system for explaining a fifth embodiment of the speech synthesis method according to the block diagram Figure 10 the present invention showing the arrangement of a speech synthesis system because EXPLANATION OF REFERENCE NUMERALS
Reference Signs List 10 prosody control unit 11 unit selection unit 12 speech unit storage unit 13 phase randomization processing unit 14 unit connection unit 15 synthesis filter unit 16 phase randomized speech unit storage unit 17 switching unit Reference Signs List 20 FFT processing unit 21 Phase conversion unit 22 IFFT processing unit 23 Noise signal generation unit 24 FFT processing unit 25 Band selection unit 100 Prosodic information 101 Prosodic control speech unit 102 Phase random prosodic control speech unit Piece 103: Phase random synthesized speech signal 104 ... Voice unit 105 ... Synthesized voice signal 111 ... Prosody control residual speech unit 112 ... Phase random prosody control residual speech unit 113 ... Phase random residual signal 114 ... Residual signal 120 phase random speech unit 200 unit waveform 201 unit waveform Fourier coefficient 202 phase random unit waveform Fourier coefficient 203 phase random unit waveform 2 4 ... noise signal 205 ... noise signal Fourier coefficients

Claims (8)

Selecting a speech unit corresponding to the specified prosody information from a speech unit dictionary consisting of a plurality of speech units;
Generating a synthesized speech signal by connecting the selected speech units;
Performing a process of randomizing the phase of the voiced sound of the synthesized voice signal. The step of performing a process of randomizing the phase of the voiced sound of the synthesized speech signal is performed by performing a Fourier transform on a noise signal and a phase of a Fourier coefficient obtained by performing a Fourier transform on a unit waveform corresponding to the voiced sound of the speech unit. 2. The speech synthesis according to claim 1, wherein a process of replacing the phase of the obtained Fourier coefficient or adding noise to a phase of a Fourier coefficient obtained by performing a Fourier transform on a unit waveform corresponding to a voiced sound of the speech unit is performed. Method. Performing a process of randomizing the phase of the voiced sound of the synthesized speech signal,
3. The speech synthesis method according to claim 1, wherein a process of randomizing a phase of a voiced sound included in the selected speech unit is performed before connecting the selected speech unit. The voice synthesizing method according to any one of claims 1 to 3, wherein the step of performing the process of randomizing the phase of the voiced sound of the synthesized voice signal performs the process of randomizing the phase of a specific band of the voiced sound. . 5. The speech synthesis method according to claim 4, wherein a process of randomizing a phase of a specific band of the voiced sound is controlled for each speech unit. 2. The speech synthesis method according to claim 1, wherein the step of performing the process of randomizing the phase of the voiced sound of the synthesized speech signal randomizes a phase of a speech unit included in the speech unit dictionary in advance. 3. A speech unit storage unit that stores a speech unit dictionary composed of a plurality of speech units,
A unit selection unit for selecting a speech unit corresponding to the prosodic information input from the speech unit dictionary,
Means for randomizing the phase of the voiced sound contained in the selected speech unit,
A prosody control unit that controls the prosody of the selected speech unit according to the prosody information,
A speech unit for connecting speech segments whose prosody is controlled to generate a synthesized speech signal. A process of selecting a speech unit corresponding to prosodic information from a speech unit dictionary composed of a plurality of speech units stored in advance,
Processing to randomize the phase of the voiced sound included in the selected speech unit;
A process of controlling the prosody of the selected speech unit according to the prosody information;
A speech synthesis program for causing a computer to perform a process of connecting synthesized speech units with controlled prosody and generating a synthesized speech signal.

Images