JPS617899A - Multipulse type encoder/decoder - 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 multi-pulse encoding/decoding device, and more particularly to a multi-pulse encoding/decoding device equipped with pitch prediction means.

[Prior art]

The sound source information of the input speech Iba code is encoded as a pseudo vocal cord wave modeled by multiple impulse sequences (multipulses) together with other necessary speech data and transmitted to the synthesis side.The synthesis side decodes these and generates the input speech. Multi-pulse encoding/decoding devices for reproducing signals have recently become well known.

This pseudo vocal cord wave converts the input audio signal into LPC (Lin-ea
It corresponds to the residual waveform of the PC coefficient as the spectral envelope information obtained by analyzing the r Prediction Coefficienζ linear prediction coefficient and the residual waveform as the sound source information.
In decoding, this residual waveform is expressed as a plurality of impulse sequences, so-called multipulses, for each analysis frame. Such a multi-pulsed sound source basically retains the waveform information included in the sound source information, and since the multi-pulsed sound source waveform has a simpler shape than the voice waveform, it can be transmitted with fewer messages. It has characteristics.

For such a multi-pulse encoding/decoding device,
Furthermore, devices incorporating pitch prediction analysis means for input audio signals have recently come into use.

This is because the multipulses obtained for each analysis frame generally have repeatability that repeats over a pitch period, so a pitch search is performed between the analysis frames and pitch prediction analysis is performed to utilize this pitch information. The aim is to reduce the data bit rate required for transmission.

FIG. 1 is an explanatory diagram of pitch prediction analysis effects for explaining the effects of pitch prediction analysis.

FIG. 1 shows an example of sound source pulses, in which two pulse groups P1 exist with a pitch period T in an analysis frame S.
and P2, and each pulse group has a plurality of pulses arranged in approximately the same time series in the positive and negative directions with respect to the time axis t. Usually, the pulse group P2 that occurs after the pitch period T has a level, fluctuation, etc. Except for the above conditions, the pulses exist as a pulse group arranged in the positive and negative directions in substantially the same time sequence as each of the six pulses of the pulse group P1. Figure 1 shows an example where there are 21 fFil of such pulse groups in the analysis frame S, but generally speaking, there are multiple such pulse groups for each pitch period, each in almost the same time sequence. exist. Therefore, in the example of FIG. 1, information regarding the first pulse group P1 and pitch prediction analysis information regarding the pitch period and pitch gain obtained through pitch prediction analysis of the input audio signal on the coding side for each analysis frame are encoded. It is possible to transmit the sound source waveform information by transmitting it from the encoding side to the decoding side, and this means that even if the sound source pulse is multipulsed, the pitch period is preserved almost the same, so it can be used in the same way. has been done. The pitch gain described above corresponds to the time ratio between the pitch period and the analysis frame that should occur repeatedly in the same seven analysis frames, and pitch prediction can be performed for each analysis frame using the pitch period and pitch gain. A multi-pulse 2 encoding/decoding device that includes pitch prediction analysis can significantly reduce the transmission data bit rate.

However, in this type of conventional multi-pulse encoding/decoding device that performs pitch prediction analysis, the reproduction quality of the input audio signal to be reproduced on the decoding side tends to deteriorate due to the following various reasons. There are drawbacks.

In other words, the pitch period extracted by pitch prediction analysis is generally rarely an integer fraction of the analysis frame, and furthermore, the effects of fluctuations in the pitch period and temporal changes in the spectrum are also superimposed on this. This is hindered by various conditions such as the fact that sound reproduction often occurs, and deterioration of the reproduced sound quality is unavoidable.

[Purpose of the invention]

An object of the present invention is to eliminate the above-mentioned drawbacks, and to provide a multi-pulse encoding and Chinese encoding device having a pitch prediction means, which is capable of encoding an input speech signal by providing means for utilizing pitch prediction analysis in the multi-pulse determination process. and decoding, which can significantly improve playback quality deterioration.
An object of the present invention is to provide a Kurusu encoding/decoding device.

[Structure of the invention]

The apparatus of the present invention is a multi-node (multiple pulse) type encoding/decoding apparatus having a pitch prediction means, in which an input signal waveform to be searched for each analysis frame in a multi-noise determination process and an input signal waveform of a synthesis filter are used. Maximum cross-correlation value with: first pulse search in which the pulse corresponding to the maximum absolute value among the extremely thick values remaining after removing the series corresponding to the pitch period from the series series is determined as the sound source representative pulse Means H: Search for the maximum absolute value pulse in a time period of one pitch period from the oldest sample in the analysis frame, or use the maximum absolute value pulse obtained by repeating this search a preset number of times as the representative pulse of the number of tones. and a second pulse search means that searches for the maximum absolute value pulse in a time corresponding to a preset number of pitch periods from the most past sample in the analysis frame, or executes this search for a preset multiple line return. a third pulse search means that determines the pulse with the maximum absolute value obtained from the pulse as the representative pulse of the sound source; a fourth pulse search means for determining the maximum absolute value pulse obtained by searching for the maximum absolute value pulse at the time of , or by repeatedly performing this search a plurality of preset times as the sound source representative pulse; a fifth pulse search means for determining the pulse with the maximum absolute value obtained by extending the pulse search means of No. 4 to the entire analysis frame as the sound source representative pulse; and the second to fifth pulses. a sixth pulse search means for determining a sound source representative pulse by repeating the search means a set number of times corresponding to the state of the pulse in the analysis frame; comprising any of the following means:

〔Example〕

Next, the present invention will be described in detail with reference to the drawings.

FIG. 2 is a block diagram showing the configuration of the encoding side (analysis side) in the first embodiment of the present invention, and FIG. ), which utilizes the first pulse search means.

The configuration on the encoding side shown in FIG. 2(A) is LPC (Lin
ear Prediction Coefficien
t, linear prediction coefficient) analyzer 1, encoder 2, cross-correlation function calculator, autocorrelation function calculator 4, pinch prediction analyzer 5
, an encoder 6, a multipulse searcher (1) 7, an encoder 8, a multiplexer 9, etc., and a demultiplexer 1 on the decoding side shown in FIG. 2(B).
0, decoders 11, 12, 13, pitch synthesizer 14, L
It is configured to include a PG synthesizer 15, an LPF 16, and the like.

The encoding side shown in FIG. 2(A) is as shown in FIG. 4(A), which will be described later.
In both the second embodiment shown in FIG. 1 and K, information regarding the pitch period and pitch gain for each analysis frame obtained by pitch prediction analysis is used in the multipulse determination process, and the sound source pulse train that should be the optimal multipulse is not selected.

Regarding the usual method of obtaining a sound source pulse train that should be a multi-pulse, it is well known that A-b-8 (Analysis-by-8ynth) by B. S. Atal et al.
esis) processing method and the correlation function processing method by Obama, Mikihiro, and Ono.While the former uses spectral domain evaluation, the latter uses cross-correlation between the input audio signal and the impulse response bonne of the synthesis filter in the analysis stage. This is a correlation region evaluation performed using Rahi and Impulse Bonne autocorrelation, etc., and is also well known as 7 forward processing, which is more efficient than A-b-5 processing. This embodiment also performs this forward processing using pitch prediction analysis.

FIG. 3 is a multi-pulse determination calculation block diagram for explaining the contents of the basic calculation algorithm for multi-pulse determination by forward processing.

The input audio signal waveform S (Z
) performs the convolution integral S (Z) by passing the transfer function W (Z) through a noise filter. The transfer function W (Z) is obtained by LPC analysis obtained through LPG analysis of the input audio signal, and the perceptual weighting of its order is set using coefficients, so that the input audio signal is Auditory weighting is applied, and then H
(Z) Perform cross-correlation with the impulse response of rice W (Z).

H (Z) (Z) is the result of convolving the sound path filter's transfer function H with the noise filter's transfer function W (Z) and giving it an auditory weight. A cross-correlation function can be calculated using the perceptually weighted impulse response of the vocal tract filter and the input speech signal.

Further, in the autocorrelation calculation, the weighting of the vocal tract filter is performed by calculating the autocorrelation function of the impulse response.

Using the cross-correlation function and autocorrelation function for each analysis frame obtained in this way, the calculation shown in the following equation (1) is executed to determine the multipulse.

Each symbol in formula (1) indicates the following content.

mi is the time position from the frame end of the first pulse in the analysis frame, gl is its amplitude, ψhx(mi)
is the cross-correlation function at time delay mi, gt is the amplitude of the pulse within the analysis frame, Rhh (1 mz-mil
) is an autocorrelation function. As is clear from equation (1), the amplitude gi(mi) is obtained by finding the difference between the cross-correlation functions, and if such a pulse is generated at the previous time position m1, the amplitude g1(mi) is determined as the optimal one. I can do it. In other words, by focusing on a certain sound source pulse and calculating its amplitude using equation (1) at various time positions, the pulse with the maximum absolute value is obtained as the pulse that is most similar to the sound source pulse, and this operation is repeated to obtain multiple pulses. The normal 7-year-old word multipulse forming means obtains a sound source pulse and uses it as a multipulse, and when a pitch prediction means is superimposed on this, a method as shown in FIG. Although the sound source pulses are encoded and transmitted to the decoding side, such encoding/decoding has the drawbacks mentioned above.

Returning to FIG. 2 (A) again, the explanation will be continued. An input audio signal input via an input terminal 1001 is supplied to an LPC analyzer 11, a cross-correlation function calculator 3, and a pitch analyzer 5.

The LPG analyzer 1 analyzes the input audio signal for each analysis frame.
It is quantized as a digital quantity with a preset number of bits, and is then analyzed by LPC to obtain a p-order parameter (partial autocorrelation coefficient) as an LPC coefficient, which is then sent to the encoder 2 via an output line 101.

The encoder 2 quantizes the input LPC coefficients, encodes them, and sends them to the output line 201. The encoder 2 also decodes the quantized LPC coefficients and performs perceptual weighting of the vocal tract filter to obtain an impulse response and output it. line 20
Send to 2.

The encoded LPC coefficients thus sent out to the output line 201 are supplied to the multiplexer 9, and also to the output line 201.
The impulse response sent to 2 is supplied to a cross-correlation calculator 3 and an autocorrelation function calculator 4.

The cross-correlation function calculator 3 calculates a cross-correlation function by convolving the impulse response and the input audio signal supplied in this way, and sends the cross-correlation function to the multi-pulse searcher (1).
Send on 7.

The autocorrelation-number calculator 4 calculates the autocorrelation function of the input impulse response and converts it into a multipulse searcher (
1).

In multi-pulse search that does not include pitch prediction, the cross-correlation function and autocorrelation function thus supplied to the analysis frame are used to find a predetermined number of sound sources based on the calculation using equation (1). The basic multi-pulse generation means is to obtain a pulse train and output the amplitude and position information of these pulses. However, in the multi-pulse determination process including pitch prediction in this embodiment, the pitch prediction analyzer 5 Using the obtained pitch prediction information, the multi-pulse searcher (1) 7 removes pulses that are more expected to have repeatability corresponding to the pitch period, and extracts the sound source pulse, the pitch period as pitch prediction information, and the pitch gain. L as spectral envelope information with only sound source information
It is transmitted to the decoding side together with the PG coefficient data, and encoding and decoding are performed that greatly suppresses deterioration of reproduced sound quality.

That is, the pitch prediction analyzer 5 utilizes an autocorrelation function calculation circuit, etc., and calculates the pitch period of the sound source pulse sequence based on the principle that the autocorrelation function calculated for each analysis frame takes a maximum value at a delay time equal to the pitch period. Predictively analyze.

The pitch period obtained in this way is sent to the encoder 6 as pitch period prediction information together with the pitch gain, which can be said to be a kind of comparison amount with the analysis frame period, and after being truncated and encoded, it is sent to the multiplexer 9 and the multipulse searcher ( 1) Supplied to 7.

The multi-pulse searcher (1) 7 utilizes the cross-correlation function xl and the pitch period prediction information (xl, autocorrelation function, and pitch period prediction information) for each analysis frame supplied in this way, and calculates the search that should be performed on a representative multi-pulse as follows. Search and determine the representative pulse.

That is, the multi-pulse searcher (1) 7 calculates the sound source pulse train using the cross-correlation function and the auto-correlation function for each analysis frame according to equation (1). What should be searched for is the output of the cross-correlation aX output unit 3, that is, the maximum sequence of cross-correlation values between the input signal waveform and the impulse response of the synthesis filter received from the encoder 2. In this embodiment, the pitch prediction analyzer 5 receives the data via the encoder 6.
The sound source pulse search using information regarding the pitch period and pitch gain is performed as follows.

That is, after removing the local maximum value series corresponding to pitch synchronization from the local maximum value sequence to be searched within the analysis frame, the sound source pulse corresponding to the maximum value among the remaining local maximum values is used as the sound source representative pulse representing the sound source. Determine as. If there is pitch prediction analysis information including the sound source representative pulse determined in this way, pitch period, and pitch gain, multi-pulse synthesis can be easily performed on the decoding side as described later. The encoder 8 encodes and decodes the representative pulses of the sound source and sends them to the multiplexer 9. The multiplexer 9 multiplexes all inputs in a splitting manner and sends them through the transmission line 901 as shown in FIG. 2 (B). ) is transmitted to the decoding side shown in FIG. .

Figure 2 (B) Demultiplexer 1 on the decoding side shown by K
0 demultiplexes the data transmitted from the code side, the sound source representative pulse is sent to the decoder 11 via the Deba line 111, and the pitch prediction analysis information/0 is sent to the decoder 12 via the output line 121. Also, the LPC coefficient is 13
1 to a decoder 13 and decoded to output lines 112, 122 and 13, respectively.
2 is output.

The pitch synthesizer 14 includes a memory circuit 141 and a pitch synthesizer 1.
42, and stores the sound source representative pulses and pitch prediction analysis information input via the output lines 112 and 122 in the analysis frame frequency memory circuit 141 and supplies them to the pitch synthesizer 142.

The pitch synthesizer 142 calculates and reproduces a sound source pulse train as a multi-pulse based on the pitch period and pitch gain based on the pitch prediction analysis information inputted via the memory circuit 141 and the sound source representative pulse, and outputs the sound source pulse train as a multi-pulse to the LPC synthesizer 1.
Supply to 5.

LPC synthesizer 1 having an all-pole digital filter configuration
5 from the decoder 13 via the output line 132
Receives the C coefficient, uses it as a filter coefficient, uses the multipulse received from the pitch synthesis filter as a filter driving sound source, reproduces the input audio signal, converts it to D/A (digital/analog), and then converts it to L P F (Low Pa5s).
F1lter) l6 and sends a predetermined low frequency component to output line 161.

In this way, it is possible to realize a multi-pulse encoding/decoding device that utilizes pitch prediction analysis in the sound source pulse determination process and significantly improves the reproduced sound quality.

FIGS. 4(A) and 4(B) are block diagrams showing the configurations of the encoding side (A) and the decoding side according to the second embodiment of the present invention. It is something that can be used.

4(A) is the same as the one with the same symbol on the encoding side of the first embodiment shown in FIG. 2(A), except for the multi-pulse searcher (2) 17, and FIG. ) shown in the first
The only difference is that a pitch synthesis filter 18 is arranged in place of the pitch synthesizer 14 on the decoding side of the embodiment of , since other items with the same symbols have the same contents, so a detailed explanation of these same contents will be omitted. do. The multi-pulse search device (1) 17 in FIG. 4(A) uses any of the five pulse search means 2nd to 6th to search for a sound source representative pulse, which can also be called a representative multi-pulse. It is used to search and decide.

First, the second pulse search means searches for the maximum absolute value within a time period of one pitch period from the most past sample in the analysis frame, and calculates this value by g as shown in equation (1).
t (mi), or the maximum absolute value returned by 4 times set in advance is set as gz(mi
)- and set this as sound source pulse information together with pitch prediction analysis information and LPC coefficient data as shown in Figure 4 (B).
2: Multi-pulses are synthesized by the sending pitch synthesizer 18 on the decoding side shown, and everything else is as shown in Fig. 2 (A) and (B).
) The input audio signal is reproduced in the same manner as in ).

In this case, since it is assumed that a multi-pulse search is performed based on the absolute maximum value search once or several times within the time of the most past one pitch period for each analysis frame, the chronicity side can also be used as shown in Figure 2 ( As shown in B), it is possible to perform pitch synthesis, that is, multi-pulse reproduction, for each input of the maximum absolute value pulse set as the sound source representative pulse, without storing analysis frames and then performing pitch synthesis.

Next, the third pulse search means expands the search range in the second pulse search means by a preset number of cycles from the most past sample, and searches for the maximum absolute value within this time. Alternatively, the pulse obtained by repeating this search a predetermined number of times is set as gt(mi) shown in equation (1), and this is sent to the decoding side as the sound source representative pulse. By expanding the search time range in this way, it is possible to increase the certainty of the maximum absolute value to be searched.

The fourth pulse search means further expands the search range, and searches for the pulse with the maximum absolute value in a time range of N multiple pitch periods from the most recent sample of the analysis frame, or sets this search in advance. The pulse obtained by repeating this several times is sent to the decoding side as a representative pulse of the sound source.

The fifth pulse search means further expands the search time range in the second to fourth pulse search means and searches for the maximum absolute value in the entire seven analysis frames. The pulse search means are the second to fifth pulse search means.
This is repeated several times to search for the absolute maximum pulse, while carrying out the examination means one after another. 2nd to 6th above
Which pulse search means will yield the desired pulse with the maximum absolute value depends on the characteristics of the mutual and autocorrelation functions for each analysis frame, the characteristics of the input audio signal, and the quality of the reproduced sound, as shown in Figure 1. For relatively clear sound reproduction, it is easy to ensure relatively good reproduction sound quality by using the second pulse search means, and generally, the second to sixth pulse search means having different search time regions are used to reproduce the reproduced sound. , and can be used as desired, depending on conditions such as processing time.

In addition, on the decoding side shown in FIG. 4(B), the pitch synthesis filter 18 is used based on the multi-pulse information obtained by any of the second to sixth pulse search means and the pitch prediction analysis information including the pitch period and pitch gain. The multi-pulse sequence is reproduced by

The pitch synthesis filter 18 includes a delay circuit that can variably set the amount of delay in accordance with the pitch period, a transversal filter that receives the output of the delay circuit, and adds the output of this transversal filter and the input to the delay circuit. The basic configuration is an adder circuit that supplies the sound source representative pulse to the delay circuit, and generates a plurality of impulse sequences arranged one after another in the pitch period, that is, multipulses, which are sent to the LPC synthesizer 15 as sound source information. Send. Note that the tap coefficients of the transversal filter included in the pitch synthesis filter 18 described above are set for each analysis frame in accordance with pitch prediction analysis information input via the decoder 12.

The present invention provides a multi-pulse encoding/decoding device that performs pitch prediction analysis, and improves playback sound quality by encoding and decoding an input audio signal using pitch prediction analysis information in the multi-pulse determination process. The basic feature is that improvements have been made, as shown in Figure 2 (A).
, (B) and the first and second embodiments of the present invention shown in FIGS. 4(A) and 4(B) can also be considered.

For example, in these first and second embodiments, L
Although a parameter (partial autocorrelation coefficient) is used as the PC coefficient, there is no problem in using other LPG coefficients, such as the α parameter, and the LPC synthesizer uses an all-pole filter. But this is another Non
It is obvious that the present invention can be implemented in the opposite manner by replacing it with a -Po1e type filter, etc., and all of the above can be easily implemented without impairing the gist of the present invention.

〔Effect of the invention〕

As explained above, according to the present invention, in a multi-pulse encoding/decoding device that performs pitch prediction analysis, pitch prediction analysis means is used in the multi-pulse determination processing process to determine pitch period and analysis frame. This has the effect of realizing a multi-pulse encoding/decoding device that can significantly improve the deterioration of reproduced sound quality due to irregular ratios, pitch cycle fluctuations, temporal changes in spectrum, etc.

[Brief explanation of the drawing]

FIG. 1 is a pitch prediction analysis effect explanatory diagram for explaining the effect of pitch prediction analysis, FIG. 2A is a block diagram showing the configuration of the encoding side in the first embodiment of the present invention, FIG. (B) is a block diagram showing the configuration of the decoding side, and Fig. 3 is a multi-pulse decision calculation block diagram for explaining the contents of the basic calculation algorithm for multi-pulse decision by forward processing. Block 1''4 shows the configuration of the encoding side of the second embodiment of the invention, and FIG. 4(B) is a block diagram showing the configuration of the first decoding side of the second embodiment of the invention.1. ... LPC analyzer, 2 ... Encoder, 3 ... Cross correlation function calculator, 4 ...
Autocorrelation function calculator, 5...Pitch prediction analyzer, 6...Encoder, 7...Multi-pulse search 6(1), 8...・Encoder, 9...
・・Multiple brephser, lO ・・Demultiplexer, 11.12.13 ・・・Decoder, 14 ・・
... Pitch synthesizer, 15 ... LPC synthesizer, 16 ... LPF, 17 ... Multi-pulse searcher (2), pitch synthesis filter, 141 ...・
...Memory circuit, 142...Pitch synthesizer. A phase # waveform 5 (Z): Input audio signal liquid form W (Z): Noise filter A thick function H (Z):
Vocal tract filter fall 1 function Z: exP (j included) = 634

Claims (1)

[Claims] A multi-pulse encoding/decoding device having a pitch prediction means, which calculates the pitch from the maximum value series of the cross-correlation value between the input signal waveform and the impulse response of the synthesis filter to be searched for each analysis frame in the multi-pulse determination process. a first pulse search means for determining, as a sound source representative pulse, a pulse corresponding to the maximum absolute value among the maximum values remaining after removing the series corresponding to the period; a second pulse search means for determining the maximum absolute value pulse obtained by searching for the maximum absolute value pulse in a time corresponding to the pitch period or repeating this search a plurality of times as the sound source representative pulse, and an analysis frame; Search for the maximum absolute value pulse in a preset number of pitch periods from the oldest sample in the sample, or determine the maximum absolute value pulse obtained by repeating this search a preset number of times as the sound source representative pulse. and a third pulse search means for searching for N times the pitch period (from the most past sample in the analysis frame).
(N is a positive integer greater than 2, including 2), or the maximum absolute value pulse obtained by repeating this search a preset number of times is determined as the sound source representative pulse. a fourth pulse search means, and a fifth pulse search means for determining the maximum absolute value pulse obtained by extending the second to fourth pulse search means to the entire analysis frame as the sound source representative pulse; , a sixth pulse search means for determining a sound source representative pulse by repeating the second to fifth pulse search means a set number of times corresponding to the state of the pulse in the analysis frame; Alternatively, a multi-pulse type encoding/decoding device comprising any one of second to sixth pulse search means to encode and decode an input audio signal.