JPS60237500A - Multipulse type vocoder - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] (Technical field) The present invention relates to a multipulse vocoder.

(Prior art) Input audio is analyzed, spectral envelope information and sound source information that constitute the audio information of this input audio signal are extracted on the analysis side, and these audio information are sent to the synthesis side via a transmission path. Vocoders for reproducing input audio signals are well known.

The above-mentioned spectral envelope information represents the spectral distribution information of the vocal tract system that generates the input speech signal, and is usually L
The number of L corresponding to n analysis order obtained by PG analysis
The sound source information is expressed by PG coefficients, such as the α parameter, and parameters, etc., and the sound source information indicates the fine structure of the spectral envelope, and is known as the so-called residual signal, which is obtained by removing the spectral distribution information from the input audio signal. It contains information about the source strength, pitch period, and voicing/voicing of the input audio signal, and it is also well known that this information is usually extracted via the autocorrelation coefficient for each analysis frame of the input audio signal. There are n.

When synthesizing input audio signals on the synthesis side of a vocoder, spectral envelope information is usually used as coefficients for an LPC synthesizer that uses an all-pole digital filter to form an approximate vocal tract system. is used as a driving sound source for this digital filter, and input audio signals are synthesized by this digital filter.

The conventional LPC vocoder obtained in this way has approximately 4
Although it is possible to synthesize speech even at a low bit rate of Kb (kilobits) or less, and it is widely used, it has the drawback that high-quality speech synthesis is difficult even at a high focus rate. The reason for this is that when modeling sound source information, for voiced sounds, the pitch period corresponding to the content is extracted and approximately represented by a single impulse train corresponding to this pitch period, while unvoiced sounds with random periods are is assumed to be a simple modeling process in which it is approximated by white noise, so it does not faithfully extract the sound source information of the input audio signal, and therefore the input audio signal contained in the sound source information This is because the analysis and synthesis of waveform information has not been performed.

A multi-pulse vocoder has recently become well known as a type of vocoder that performs waveform transmission and synthesizes input audio signals in order to improve the problem caused by non-transmission of waveforms.

FIG. 1 is a block diagram showing the basic configuration on the analysis side of a conventional multi-pulse vocoder.

The LPG synthesizer 1 is equipped with an all-pole digital filter that simulates the vocal tract, and its coefficients are based on the input audio signal x(n) (n-1, 2, 3, -
.

The sound source pulse generator 3 obtains a driving sound source sequence V(n) consisting of a plurality of impulse sequences, that is, multinoculus, from the sound source information of the input audio signal, and converts this n into an LPC synthesizer 1.
supplied as a driving sound source.

The LPC synthesizer 1 converts the input I and PC coefficients into a combination using an all-pole digital filter.
l-jl's melon erection 1 Multi I Kurusu ffimGfl
n synthesized signal '3i'' (n) e is driven as fi. In this case, the multipulse contains the waveform information of the input audio signal, and the LPG synthesizer l synthesizes the input audio signal including the waveform information. It will be done.

Next, the synthesized signal X(n) output from the LPC synthesizer 1 is subtracted by the subtracter 4, which takes the difference from the input audio signal x(n) to obtain the error e(n). Send to 5.

The auditory weighting device 5 calculates the following (1) for the error e(n).
The weighting having the characteristic W(Z) shown in the equation is performed by applying perceptual weighting using a filter, and then sending these to the squared error minimizer 6.

W(Z)=(1-5; a Z-”)/(1-set a*r
'Z'']]k-1kk-1... (1) In equation (1), ak is the LPC coefficient that should be the coefficient of the all-pole digital filter of LPC synthesizer 1, p is the order, therefore the LPG analysis order, γ is the weighting coefficient, and Z is the transfer function H in the Z-transform representation of the all-pole digital filter.
(Z-'), where λ = 2πΔTf and ΔT is the sampling rate of the analysis frame y
The pulling period, f indicates the frequency.

Further, in equation (1), the weighting coefficient γ is set in the range of 0<γ<1.

W(Z) shown in equation (1) is 1 for r=1 and γ=0
The value of r varies within the range of W(Z) = 1-p(Z), and the value of r is set to an extent that suppresses the excessive level that appears in the formant region of the frequency spectrum of error e(n). It is set within the above-mentioned range corresponding to 1, and plays the role of perceptual weighting of the signals to be synthesized, and its optimal value is usually selected in advance by an optimal auditory test.

The error e(n) weighted in this way is the drive sound source series V (n ) output from the sound source pulse generator 3.
, that is, the following (2) is sent to the squared error minimizer 6 to determine the optimal time position and amplitude of the multi-pulse.
The squared error ε is calculated using the formula, and the driving sound source sequence V(n) is selected so as to minimize ε.

ε=l: (e(n) Kaoru W(n))” ・・・－・・・・・・
(2)-1 In equation (2), the symbol box indicates the auditory weighting, the weighting of the device 5 indicates convolution integration by a filter, and N indicates the interval length for calculating the multipulse.

The above process is repeated for each pulse of the multipulse,
This is a so-called analysis-by-8 synthesis method (hereinafter abbreviated as A-b-8 method) in which synthesis by analysis is performed for each multipulse.

However, the multi-pulse vocoder has the following drawbacks.

In other words, when the number of pulses required to reproduce the synthesized sound of the desired sound quality existing within the frame is greater than the number of pulses generated as a result of analysis within the frame, the synthesized signal transmits the waveform related to the sound source information of the human voice signal. The result will not be faithfully executed, and the audio quality of the synthesized signal will deteriorate to a degree corresponding to the difference in the number of pulses described above.

In a multi-pulse vocoder, for example, the analysis period t-2
The number of sound source drive pulses that should be generated in one frame with 0m S E C is usually 4 to 1 depending on the pit rate.
When the input audio signal is a high-pitched voice with a small pitch period, such as a female voice or a child's voice, it is not uncommon for the pitch period of the sound source signal to be about 25m5EC, in this case 1. At least eight driving sound source pulses are required to be set in an analysis frame. In such a case, if the number of driving sound source pulses to be generated within the analysis frame is set to 8 or less, for example 4, a multi-pulse type vocoder that uses such driving sound source pulses will suffer from a double pitch error. A synthesized sound containing the result will be generated, and the synthesized sound quality will be significantly degraded.

Attempts have been made to introduce pitch prediction coefficients into multi-pulse vocoders in order to eliminate the above-mentioned drawbacks. The pitch prediction coefficient represents the property that speech has approximately periodicity, with the pitch period, that is, the imaging period of the vocal cords being the repetition period. The use of pitch prediction coefficients in a multi-pulse vocoder is achieved by inserting a pitch synthesizer between the sound source pulse generator and the LPG synthesizer, which determines the transfer function based on the pitch period and pitch prediction coefficients. It will be done.

FIG. 2 shows an example of a configuration on the analysis side when a pitch synthesizer is inserted into a multi-pulse vocoder. Reference numbers 1-6 in FIG. 2 may be the same as blocks with the same reference numbers in FIG. The pinch analyzer 7 calculates the pitch period T of the audio signal and the pitch prediction coefficient b l (i = Tp-j, Tp-j+1...) using an autocorrelation method or the like.
..., T,, T, +1. ...-2T, 1j-1, T, 1j). The pitch synthesizer 8 is a type of filter driven by the sound source pulse train, and has a transfer function equivalent to the pitch prediction characteristic shown in equation (3) below.

By using the pitch synthesizer 8, the sound source pulse generator 3
This means that the number of flashing pulses that can be generated can be kept to a minimum. Therefore, when the number of sound source driving pulses is fixed, the above-mentioned drawbacks are alleviated. For information on how to use pitch prediction coefficients in multi-pulse vocoders, see Obama et al., ``Multi-pulse driven speech coding method using pitch information,'' Acoustical Society of Japan, 1988 Autumn Research Presentation Proceedings.

Oct. 1983 2-2-14.

However, the introduction of pitch prediction coefficients creates a need to transmit pitch prediction coefficients and the like from the analysis side to the synthesis side. Therefore, in a multi-pulse vocoder with a constant transmission bit rate, such as the 96008P8, the coefficients are introduced by deleting the amount of information of other parameters. Therefore, the introduction of pitch prediction coefficients has the following three consequences in terms of sound quality.

■ The sound quality of synthesized speech is improved by introducing pitch prediction coefficients.

■ The pitch prediction effect due to the introduction of the pitch prediction coefficient and the deterioration in sound quality due to the reduction of parameters other than the pitch prediction coefficient cancel each other out, and there is almost no effect of improving sound quality.

■ The effect of sound quality deterioration due to the reduction of parameters other than the pitch prediction coefficient is more significant than the pitch prediction effect, and rather leads to deterioration of the synthesized sound.

Conventionally, multi-pulse vocoders that have introduced pitch prediction coefficients have uniformly performed pitch prediction, which has the disadvantage of sometimes deteriorating the synthesized sound quality, as described in (■) above. .

(Object of the Invention) The object of the present invention is to eliminate the above-mentioned drawbacks and provide a multi-pulse vocoder having the best sound quality encoding means in a multi-pulse vocoder with a predetermined transmission bit rate. There is a particular thing.

(Structure of the Invention) A specific example of the multi-pulse vocoder of the present invention performs LPC analysis and extracts input speech signals for each analysis frame.
Using the PC coefficients as spectral envelope information, together with this spectral envelope information, the sound source information constituting the audio information of the input audio signal is analyzed. A plurality of impulse sequences having generation time positions and amplitudes corresponding to the characteristics of this sound source information are generated for each frame. A multi-pulse vocoder is used to analyze and synthesize the input audio signal.
and comparing and determining the pitch prediction effect with the influence of other parameters that are reduced when pitch prediction is performed, using the pitch period and pitch prediction coefficient extracted for each analysis frame of the input audio signal, If the pitch prediction effect is large, the pitch period and pitch prediction coefficient are transmitted from the analysis side to the synthesis side, and if the effect is small, a multi-pulse driven sound source is used instead of the period and/or pitch prediction coefficient. The device is constructed with means for transmitting more information.

(lol example) Next, the present invention will be described in detail with reference to the drawings.

FIG. 3 is a block diagram showing an embodiment of the analysis side of a multi-pulse vocoder according to the present invention, and FIG. 4 is a block diagram showing an embodiment of the synthesis side of the multi-pulse vocoder according to the present invention.

The analysis side of the multi-pulse vocoder according to the present invention shown in FIG. 3 consists of an LPC synthesizer 1. LPC analyzer 2, sound source pulse generator 3. Subtractor 4. Auditory weighting unit 5. Square error minimizer 6. Pitch analyzer 7° pitch synthesizer 8. Switcher 9.
Quantizer 10. Quantization decoder (1) 11. Pitch prediction control signal generator 12° quantization decoder (2) 13,
It consists of 14 multiplexers. In the figure, reference numbers 1 to 8 are the same as blocks with the same reference numbers in FIG.

The input audio signal is input to the LPC analyzer 2 through the waveform input terminal 2001e. Subtractor 4. The signal is supplied to the pitch analyzer 7. LP
The C analyzer 2 performs LPG analysis on the input audio signal using LPC analysis using techniques such as an autocorrelation method, and calculates LPG coefficients. The LPG analyzer 2 further outputs the calculated LPG coefficients to the quantization decoder (2) 13. Quantization decoder (2)
13 quantizes the supplied LPC coefficients in order to express them with a required number of bits, for example, about 48 bits. The quantized LPC coefficients are supplied to multiplexer 14 via output line/131. The quantization decoder (2) 13 further decodes the quantized LPG coefficients and outputs the output line 13.
2 to the LPC synthesizer 1 and perceptual weighter 5. The pitch analyzer 7 calculates the pitch period and pitch prediction coefficient of the input audio signal, for example via an autocorrelation coefficient. FIG. 5 is a waveform diagram for explaining the operation of the pitch analyzer 7. In FIG. 5, an autocorrelation coefficient sequence 501 is a coefficient sequence calculated from the input audio signal using the following equation (4).

Point 502 is pitch prediction coefficient b corresponding to pitch period Tp
The Tp-11 point 504 is the pitch pre-1fltl coefficient bTp4.corresponding to the period Tp-2. ...-..., is shown. or,
Point 512 is the pinch prediction coefficient bT corresponding to the period Tp+1
, +1° point 513 indicates the pitch prediction coefficient bT, +2# . . . -0 corresponding to the period Tp+2. The lock pitch period Tp is within the pitch period distribution range, for example 2.5m5EC~
There is a dizziness associated with the maximum value of the autocorrelation coefficient sequence 501 at 15m5EC (equivalent to a delay of 20 to 120 taps with 8 taps/pull).

Returning to FIG. 3 again, the pitch analyzer 7 calculates @n,
The fc pitch period data and pitch prediction coefficients are output to a quantization decoder (1) 11 and a pitch prediction side signal generator 12. The quantization decoder (1) 11 quantizes the supplied pinch period data and pitch prediction coefficient into a finite number of bits and outputs it to the multiplexer 14. Furthermore, the quantization decoder (1) 11 decodes the quantized pitch period data and pitch prediction coefficients and outputs them to the pitch synthesizer 8. The pitch prediction control signal generator 12 receives pitch period data Tp and pitch prediction coefficient bT supplied from the pitch analyzer 7.
, it is determined whether the following equation (4) holds true.

Tp-LE-TR-AND-b,,・G'l'-bR
(4) In equation (4), T is a constant set to, for example, 63, and bR is set to, for example, 07. 2. The sound quality of T&bB etc. has been experimentally evaluated. (
4) In equation 4, the pitch period T is within the preset range T.

In addition, the pitch prediction coefficient bTp is a preset value bR
This only holds true if it exceeds . If the equation (4) holds true, the pitch prediction control signal generator 12 sends a control signal to connect the switch 9 to the pitch synthesizer 8 side, and if the equation (4) does not hold, the pitch prediction control signal generator 12 connects the switch 9 to the pitch synthesizer 8 side. Generates a control signal to be connected to the sound source pulse generator 3 side. Therefore, if the above equation (4) does not hold, the processing structure of this embodiment matches that of the analysis side of the conventional multi-pulse vocoder that does not perform pitch prediction, as shown in FIG. If this holds true, the processing structure of this embodiment matches that of the analysis side of the conventional pitch prediction type multi-pulse vocoder shown in FIG. 2IC. The pitch prediction control signal generator 12 further outputs two short signals equivalent to the control signals for the switch 9 to the multiplexer 14, the quantizer 10, and the excitation pulse generator 3.

Depending on the connection condition of the switch 9, the square error minimizer 6° music pulse generator 3. Pitch synthesizer8. LPC synthesis instrument crusher 4. The loop based on the A-'b-B method in which perceptual weighting is performed determines an optimal sound source pulse sequence while appropriately performing pitch prediction. The sound source pulse sequence is a plurality of predetermined impulse sequences (multipulses) having generation time positions and amplitudes corresponding to the characteristics of the sound source information,
Direct ICU Ki 11 Hyakuba 117 '1 ``Tade Shout''l
If? The benefit pressure is given and output to the source 1^.

Of course, if the pitch pre-planning is not carried out, the number of multi-pulses can be increased by the same amount as the hitch data.

The quantizer 10 quantizes the multi-pulse train into a finite number of bits (the number of bits changes depending on the control signal mentioned above) and outputs it to the multiplexer 14. quantized multi-pulse, quantized decoder (1), quantized pitch period 2 pitch prediction coefficient supplied from the pitch prediction control signal generator 12; A binary signal equivalent to the control signal for the switch 9 and the quantization decoder (2) 1
It multiplexes the quantized LPC coefficients supplied by the third unit 9 and outputs it to the data output terminal 141. Of course - without directly transmitting the two short signals equivalent to Ml memorized control information, it is possible to use the n sign bits assigned to the pitch period or the pitch prediction coefficient when the two short signals mean that no pitch prediction is carried out. By setting a specific code, information that does not involve pitch prediction can be transmitted.

For example, the pitch period may be set to 0'' and transmitted.

The synthesis side shown in FIG. 4 synthesizes input audio signals based on the data transmitted to the analysis side via the analysis side data output terminal 141 and the data input terminal 15011, and the demultiplexer 16° Decoder (1) 17
, decoder (2) 18 , decoder (3) 19 .

Pitch synthesizer 20. Switcher 23. LPC synthesizer 21゜L
PF (Low Pa5s Filtcr) and the like.

The demultiplexer 16 restores each food data input through the data input terminal 1501 to the state before multiplexing of the multiplexer 14, and the heel multi-noise data is output on the output line 1.
61 to the decoder (1) 17, and the LPC coefficient data to the decoder (2) 18 via the output line/162.
Pitch period and pitch prediction coefficient data are output lie/16
3 to the decoder (3) 19, and two short signals equivalent to the control signals for the switch 9 described above are supplied to the switch 23 via the output line 164e. The encoder decodes the data and then sends it to the output lines 171, 181, and 191.

The pinch synthesizer 20 has the same structure as the pitch synthesizer 8 on the analysis side in FIG. The switch 23 switches the pitch synthesizer 20 by two short signals supplied from the output line 164.
The output pulse 1 of LPc can be supplied to the LPc combiner 21, or the multipulses output from the decoder (1) 17 can be directly converted into LPC.
It controls whether or not it is supplied to the synthesizer 21.

The LPG synthesizer 21 uses the input multi-pulse to which the pitch prediction is appropriately added as sound source information for the drive sound source of the p-order all-pole digital filter, and also inputs it through the output line y18 it. Using this LPG coefficient data as the coefficient of the all-pole digital filter,
Controls the PC synthesis filter to synthesize the input audio signal, and sends it to the LPF 22 via the Lk output line 211.
A predetermined one-band filtering is performed and the resultant signal is sent to the output line 221 as an analog synthesized voice.

In addition, on the analysis side, without directly transmitting the 2-short signal equivalent to the control signal of the switch 9, for example, the pitch reserve loss 11 coefficient can be calculated.
When transmitting information meaning that pitch prediction is not performed by means such as setting it to 0'', the decoder (3) 19 on the synthesis side can easily decode the pitch prediction coefficient to 101. It is clear that the switch 23 and the output line 164 are unnecessary in this case.

Furthermore, in the embodiments of the present invention shown in FIGS. 3 and 4, parameters are used as LPC coefficients, but other LPG coefficients such as the α parameter may also be used. It is clear that the encoder and multiplexer, the decoder and the demultiplexer can be implemented in the same way as integrated components, and the LPC synthesis filter can also be implemented as a non-all-pole type. It is also clear that the implementation can be carried out in almost the same way even if a polar digital filter or the like is substituted.

(Effects of the Invention) As explained above, the present invention enables a multi-pulse vocoder to determine the effectiveness of allocating bits to pitch periods and pitch prediction coefficients under conditions where the transmission bit rate is constant. By allocating bits, the synthesized sound quality can be improved the most.

[Brief explanation of drawings]

Figure 1 is a block diagram showing the basic configuration of a multi-pulse vocoder that does not perform conventional pitch prediction, and Figure 2 is for explaining the basic configuration of a multi-pulse vocoder that performs conventional pitch prediction. FIG. 3 is a block diagram showing an embodiment of the analysis side of the multi-pulse vocoder according to the present invention, and FIG. 4 is a block diagram showing an embodiment of the synthesis side of the multi-pulse vocoder according to the present invention. FIG. 5 is a waveform diagram for explaining the operation of the pitch analyzer 7. 1.21...LPC synthesizer, 2...L
PC analyzer, 3...-Sound source pulse generator, 4...
...Subtractor, 5... Auditory weighting is device, 6.
... Square error minimizer, 7 ... Pitch analyzer, 8, 20 ... Pitch synthesizer, 9.23.
...Switcher, 10...Quantizer, 11.
-...Quantization decoder (1), 12...Pitch prediction control signal generator, 13...-Quantization decoder (2), 14...・Multiplexer, 16・
...Demultiplexer, 17...Decoder (1), 18...Decoder (2), 19...
Decoder (3), 22...LPF processing)

Claims (1)

[Claims] (1) Input audio signal is analyzed by LPC (Li
near Prediction Coefficie
nt, linear prediction coefficient) is analyzed and extracted as spectral envelope information, and together with this spectral envelope information, sound source information constituting the audio information of the input audio signal is analyzed, and the generation time corresponding to the characteristics of this sound source information is determined for each analysis frame. In a multi-pulse vocoder that analyzes and synthesizes the input audio signal by expressing it using a plurality of predetermined impulse sequences (multipulses) having positions and amplitudes, an analysis frame of the input audio signal is used. a pitch analysis means for determining a pitch period and a pitch prediction coefficient; and the pitch period is within a preset range, and the pitch prediction coefficient exceeds a preset value. In this case, the pitch period and pitch prediction coefficient are transmitted to the synthesis side, and in other cases, the bits assigned to the pitch period and pitch prediction coefficient in the above case are transmitted using as multipulse encoded bits. Furthermore, a code that informs the synthesis side whether or not to transmit the pitch period and the pitch prediction coefficient is synthesized 9.
1. A multi-pulse vocoder characterized in that the analysis side has a coding control means for transmitting data to the coding control means, and the synthesis side has a decoding control means corresponding to the coding control means and a pitch synthesizer on the synthesis side. In the multi-pulse vocoder according to claim (1), the encoding control means controls all or part of the constituent bits of either the pitch period or the bits allocated to the pitch prediction coefficient. A multi-pulse vocoder characterized in that it is a means for notifying a total synthesis side whether or not to transmit a pitch period and a pitch prediction coefficient by setting the pitch period and pitch prediction coefficient to a specific code.