JPS6239760B2 - - Google Patents

Description

[Detailed Description of the Invention] (1) Technical Field of the Invention The present invention relates to a speech analysis processing method, in particular, extracting a power spectrum by Fourier transforming an input speech signal, and using the power spectrum to calculate an autocorrelation coefficient. In an audio analysis processing method that calculates linear prediction coefficients, extracts spectral envelope information of an input audio signal using the linear prediction coefficients, for example, compression or expansion is performed on the frequency domain after the Fourier transform. After performing the corresponding transformation, the linear prediction coefficient is further calculated from the autocorrelation coefficient, and the deformed spectrum envelope information obtained using the linear prediction coefficient is inversely transformed. The present invention relates to a speech analysis processing method that is capable of extracting, for example, spectral envelope information in a manner unaffected by frequency, or extracting zero frequencies and formant frequencies with high precision.

(2) Technical Background and Problems Conventionally, linear prediction coefficients have been extracted to extract parameters used in speech synthesis, speech recognition, and the like. In the above-mentioned speech synthesis and speech recognition, the spectral envelope information of the input speech signal is obtained from the above-mentioned linear prediction coefficients by, for example, treating the prediction coefficient itself as a time function and performing Fourier transform to calculate the inverse spectrum of the spectrum. , or further use the spectral envelope information to obtain formant frequencies and the like.

However, in the case of the conventionally known method for extracting spectral envelope information, there are problems such as the obtained spectral envelope information being affected by the pitch frequency of the input voice.

(3) Purpose and structure of the invention The present invention aims to solve the above-mentioned problems and to provide flexibility for future analysis processing in speech synthesis and speech analysis. The purpose is to take the content of the patent application filed earlier by one step further. Therefore, the audio analysis processing method of the present invention performs a Fourier transform on an input audio signal to convert it into the frequency domain and extract the power spectrum of the input audio signal,
In a speech analysis processing method, the power spectrum is used to calculate an autocorrelation coefficient to extract a linear prediction coefficient, and the linear prediction coefficient is used to extract spectral envelope information of the input speech signal. In addition to inserting a transform processing unit that compresses or expands the input signal in the frequency domain after Fourier transforming the input signal and before calculating the autocorrelation coefficient, the linear prediction coefficient is configured to perform an inverse transformation process to reverse the transformation performed in the transformation processing unit on the deformed spectrum envelope information obtained using the transformation processing unit, and the transformation processing unit that performs the compression or expansion is , is characterized in that the normalized value of the power spectrum P(ω) is multiplied by a coefficient μ, and logarithmic transformation is performed to take the logarithm. This will be explained below with reference to the drawings.

(4) Embodiments of the invention FIG. 1 shows an example of a conventionally known configuration for extracting spectral envelope information, and FIG. 2 shows an example of a configuration for extracting spectral envelope information according to an invention previously made by the present inventors. 3 and 4 are explanatory diagrams for explaining the prerequisite problems of the present invention, FIG. 5 is an embodiment of the configuration of the present invention, and FIG. Explanatory diagrams illustrating the advantages of the present invention using formant frequencies extracted based on spectral envelope information obtained by the configuration, Figures 7 to 12 illustrate the present invention when the coefficient μ is plotted on the horizontal axis, respectively. An explanatory diagram illustrating the characteristics of spectral envelope information obtained by.

In FIG. 1, numeral 1 is a Fourier transform processing unit that performs Fourier transform on a discrete input audio signal S(n), and numeral 2 is a square value extractor that extracts the power spectrum P(ω) of the input audio. something that extracts
3 is an inverse Fourier transform processing unit that performs inverse Fourier transform on the power spectrum P(ω) to calculate the autocorrelation coefficient R(n); 4 is a linear prediction coefficient calculation unit that calculates the autocorrelation coefficient R(n); Correlation coefficient R
(n), 5 is a Fourier transform processing unit that performs Fourier transform by regarding the linear prediction coefficient a(n) as a time function, and 6 is a square The value extractor 7 represents a reciprocal number processor. Note that the Fourier transform processing section 5, square value extraction section 6, and reciprocal number processing section 7 are as follows:
It may be considered that the spectral envelope information P(ω) of the input audio signal is extracted from the linear prediction coefficient a(n). The linear prediction coefficient calculation unit 4 may be used, for example, in (i) “Digital Signal Processing of Speech (Volume 2)” published by Corona Publishing in 1982, translated by Hisaki Suzuki, pages 165 to 165.
167 pages, (ii) IE ³ Proceeding Vol63, No.4, 1975
“Linear Prediction: a Tutorial Review” (J.
Makhoul) P566, which is conventionally known as shown in formula (37) or formulas (38a) to (38c).

When the conventionally known configuration shown in FIG. 1 is used, the following problems are involved. That is, (A) the pitch frequency of the currently input voice is (i) 62.5 or
a number of audio signals A within the frequency range of 500 Hz; (ii) a number of audio signals B within the frequency range of 83.3 to 250 Hz; and (iii) a number of audio signals within the frequency range of 62.5 to 125 Hz. C.(iv)250
If logarithmic spectral envelope information is extracted for a large number of audio signal groups D within a frequency range of 500 Hz to 500 Hz, and the root mean square of the deviation from the true logarithmic spectral envelope information of the input audio is plotted for each group, As shown in FIG. 3, the values k ₁ , k ₂ , k ₃ on the horizontal axis γ = 1.0 are preferably the same values for each audio signal group A, B, C, D, but the values shown in the diagram are The deviation has different values, such as. The value of γ will be described later, but the case of γ=1.0 corresponds to the conventional case. This indicates that the existence of the pitch frequency of the input voice causes an error in the extracted spectral envelope information, and that the extracted spectral envelope information varies in accordance with variations in the pitch frequency.

(B) Also, for each of a number of audio signals having a pitch frequency F ₀ in a range where the F ₁ /F ₀ ratio is 0.80 to 8.00 corresponding to a constant formant frequency F ₁ (500Hz), the extracted formant frequency is When plotting the degree of relative error to the true formant frequency _F1 , as shown in Figure 4,
Pitch frequency with relative error of F ₁ / F ₀ ratio of 4.00 or more
The audio signal with F ₀ should also be plotted on the error "0.00" line, but it takes a value of about ±2.50%.

As mentioned above, when the conventionally known method shown in FIG. 1 is used, the obtained spectral envelope information and the obtained formant frequency contain relatively large relative errors depending on the pitch frequency of the input audio signal. ing.

FIG. 2 shows an example of a configuration for extracting spectral envelope information using the method of the invention previously developed by the present inventors. Codes 1 to 7 and S in the figure
(n), P(ω), ^P(ω) correspond to Fig. 1, and 8
is the transformation processing section created in Fig. 2, 9 is the inverse transformation processing section, P'(ω) is the deformed power spectrum, R'(n) is the deformed autocorrelation coefficient, and a'(n) is the deformed prediction coefficient. , P'(ω) represents deformed spectrum envelope information.

In FIG. 2, the power spectrum P(ω) of the input voice is obtained by the square value extractor 2, and for example, P′(ω)=[P( ω)〕〓 ...(1) A conversion processing unit 8 is inserted which provides the following conversion. Corresponding to the value of the coefficient γ in the conversion processing unit 8, when 0<γ<1, the power spectrum P(ω) is compressed with respect to the amplitude axis, and when 1<γ, it is expanded, and -1 If <γ<0, it is compressed and the reciprocal is taken, and if γ<-1, it is expanded and the reciprocal is taken.

In the case shown in FIG. 2, the input audio signal S(n) is Fourier-transformed and the power spectrum P(ω), whose absolute value is taken, is transformed as shown in equation (1), and then, Deformed autocorrelation coefficient R'(n), deformed prediction coefficient a'(n), deformed spectrum envelope information ^P'(ω)
Then, the inverse transformation of the transformation of the above equation (1) is performed in the inverse transformation processing section 9. That is,
In the frequency domain after the Fourier transform of the input audio signal S(n) and before the inverse transform by the Fourier inverse transform processor 3, the transform shown in equation (1) is performed to obtain spectral envelope information. To extract ^P(ω), inverse transformation ^P(ω) = [^P'(ω)
〕 ^-
I am trying to do the following. Incidentally, although the amount of calculation becomes large, the transform processing section 8 may be inserted immediately after the Fourier transform processing section 1 shown in FIG.

FIG. 3 shows each audio signal group A, B,
For each of C and D, using the configuration shown in FIG. 2, the results of plotting the deviation of the spectral envelope information described above while changing the coefficient γ are shown. In the case shown in the figure, the deviations for each group A, B, C, and D are concentrated near zero near γ = 0.5, indicating that the influence of fluctuations in the pitch frequency of the input voice is absorbed. Accompanied. That is, Figure 6A is the fourth
Figure 6B is the same graph corresponding to Figure 6B.
A graph similar to that shown in FIG. 6A is shown using spectral envelope information P(ω) obtained by the configuration shown in the figure. As is clear from comparing FIG. 6A and FIG. 6B, stability is achieved when the F ₁ /F ₀ ratio is 4.00 or more, and the influence of different pitch frequencies of input audio is largely suppressed.

In the configuration shown in FIG. 2, the conversion processing section 8 performs the conversion corresponding to equation (1). The present inventors were searching for a better conversion mode, and We have found that it is preferable to perform the transformation given by Note that G in equation (2) can be considered to be for normalizing the power spectrum P(ω), and μ is an arbitrary coefficient with a positive value, and
The value 1 in the log bracket can be considered to prevent the logarithm value from taking a negative value.

FIG. 5 shows an embodiment of the present invention in which conversion is performed according to equation (2). Reference numerals 1 to 7 in the figure correspond to those in FIG. 1, 10 is a conversion processing section which corresponds to the conversion processing section 8 shown in FIG. 2, and performs the conversion according to equation (2); 2 is an inverse transformation processing section which corresponds to the inverse transformation processing section 9 shown in FIG. 2, which performs the inverse transformation of the transformation according to equation (2).

The operation according to the configuration shown in FIG. 5 is substantially the same as that shown in FIG. 2, and the operations of the conversion processing section 10 and the inverse conversion processing section 11 correspond to equation (2). Only.

The formant frequency extracted using the spectral envelope information obtained by the configuration shown in FIG. 5 and compared with that in FIGS. 6A and 6B is shown in FIG. 6C ( Note that the coefficient μ is set to a value of 10). As can be seen from FIG. 6C, the values are more stable than in the case of FIG. 6B, and it can be seen that the influence of different pitch frequencies of the input audio is suppressed.

FIGS. 7 to 9 respectively show spectral envelope information ^P(ω) for groups A, B, C, and D explained in connection with FIG. 3 based on the configuration shown in FIG.
is obtained, and the characteristics are shown when the coefficient μ at that time is changed. Also, Figures 10 to 12 are
The formant frequency is calculated using the spectral envelope information obtained by the same processing as in Figs. 7 to 9, respectively.
It shows the characteristics when F ₁ is extracted. Coefficient μ
It can be seen that the value of is preferable near the value 10.

(5) Effects of the Invention As explained above, according to the present invention, it is possible to eliminate the influence of differences in pitch frequencies of input audio signals, and the effect is particularly large in the vicinity of μ=10. Furthermore, by selecting the value of .mu., in some cases it is possible to expand and extract features due to differences in pitch frequencies of the input audio signal, improving the degree of freedom in analysis processing. Furthermore, in the present invention, the modified spectral envelope information obtained in the process of processing or the formant frequency itself determined from the modified spectral envelope information is less affected by fluctuations in the pitch frequency, so it can be used for speech synthesis and speech recognition. It is.

[Brief explanation of the drawing]

FIG. 1 is an example of a conventional configuration for extracting spectral envelope information, FIG. 2 is an example of a configuration for extracting spectral envelope information according to an invention previously made by the present inventors, and FIGS. 3 and 4 is an explanatory diagram explaining the prerequisite problem of the present invention, FIG. 5 is an example configuration of the present invention, and FIG. 6 A, B, and C are spectra obtained with each configuration shown in FIGS. 1 to 3. An explanatory diagram illustrating the advantages of the present invention using formant frequencies extracted based on envelope information, and Figures 7 to 12 are spectra obtained by the present invention when the coefficient μ is plotted on the horizontal axis, respectively. An explanatory diagram illustrating the characteristics of envelope information is shown. In the figure, 1 is a Fourier transform processing section, 2 is a square value extraction section, 3 is a Fourier inverse transform processing section, 4 is a linear prediction coefficient calculation section, 5 is a Fourier transform processing section, 6 is a square value extraction section, 7 is the reciprocal processing unit, 8 and 10 are the transformation processing units, 9 and 11 are the inverse transformation processing units, S(n) is the input audio signal, P(ω) is the power spectrum, ^P
(ω) is the spectral envelope information, P″(ω) is the deformed power spectrum, R″(n) is the deformed autocorrelation coefficient, a″(n) is the deformed prediction coefficient, ^P″(ω) is the deformed spectrum Represents envelope information.

Claims (1)

[Claims] 1 Fourier transform the input audio signal, convert it into the frequency domain, extract the power spectrum of the input audio signal, calculate the autocorrelation coefficient using the power spectrum, extract the linear prediction coefficient, and calculate the linear prediction coefficient. In a speech analysis processing method for extracting spectral envelope information of the input speech signal using prediction coefficients,
Inserting a transformation processing unit that compresses or expands the input signal in the frequency domain after Fourier transforming the input audio signal and before calculating the autocorrelation coefficient, The deformed spectrum envelope information obtained using the linear prediction coefficients is configured to undergo an inverse transformation process to perform an inverse transformation of the transformation performed in the transformation processing section, and performs the compression or expansion described above. A speech analysis processing method characterized in that the conversion processing unit performs logarithmic conversion by multiplying the normalized value of the power spectrum P(ω) by a coefficient μ and taking a logarithm thereof.