JPH11259098A - Method of speech encoding/decoding - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve speech encoding of speech quality by using an algebraic structure code book which is decreased in the number of pulse positions and pulses by lowering a rate. SOLUTION: When a speech signal is encoded by expressing it by information presenting at least a characteristic of LPC synthesis part 120, a pitch vector to drive the LPC synthesis part 120, and a driving signal consisting of noise vectors, the pitch vector is searched for from an adaptive code book 141 and also a pulse position candidate varying in pulse position according to the shape of the pitch is searched for by a vector candidate search part 142; the adaptive algebraic structure code book 143 produces a pulse sequence by arranging the pulses in the pulse position chosen by this pulse position candidate; and constitutes the driving signal of the LPC synthesis part 120 from these pitch vector and pulse sequence.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a digital telephone,
The present invention relates to a low coding rate speech encoding / decoding method used for voice memos and the like.

[0002]

2. Description of the Related Art In recent years, a CELP (Code Excited Lithium) method has been used as an encoding technique for compressing voice and musical sounds into a small amount of information for transmission and storage on a cellular phone or the Internet.
nearPrediction (MRSchroeder and BSAtal, "Code
Excited Linear Prediction (CELP): High Quality S
peech at Very Low Bit Rates, "Proc. ICASSP, pp.937
-940, 1985 (Reference 1) and WSKleijin, DJKrasin
ski et al. "Improved Speech Quality and Efficient
Vector Quantization in SELP, "Proc. ICASSP, pp.155
-158, 1988 (Reference 2)) are often used.

[0003] CELP is a coding method based on linear prediction analysis, and an input speech signal is divided into linear prediction coefficients representing phoneme information and prediction residual signals representing pitches and the like by linear prediction analysis. A recursive digital filter called a synthesis filter is formed based on the linear prediction coefficients, and the original input speech signal can be restored by inputting a prediction residual signal as a drive signal to the synthesis filter.

In order to perform encoding at a low rate, a linear prediction coefficient, which is synthesis filter information representing the characteristics of a synthesis filter, and a prediction residual signal, which is a drive signal for driving the synthesis filter, are encoded with a smaller amount of information. There is a need to. CEL
In the P system, a signal obtained by encoding a prediction residual signal is generated as a drive signal by multiplying two types of signals, that is, a pitch vector and a noise vector, by an appropriate gain and then adding the signals. A method of generating a pitch vector is described in, for example, Reference
It is described in.

[0005] In addition to the method of Reference 2, the voice onset (onset)
Although a method using a fixed code vector has been proposed, these are collectively referred to as a pitch vector in the present invention. A noise vector is usually generated by storing a large number of candidates in a random codebook and selecting an optimum one from the stored candidates. As a search method of the noise vector,
A noise vector that adds all the noise vectors to the pitch vector, passes through a synthesis filter to generate a synthesized voice signal, evaluates the distortion of the synthesized voice signal with respect to the input voice signal, and generates a synthesized voice signal with the least distortion. Is chosen. Therefore, how to efficiently store the noise vector in the noise codebook is an important point of the CELP method.

[0006] Algebraic Codebook (J-
P. Adoul et al, “Fast CELP Coding based on algebr
aic codes ”, Proc. ICASSP'87, pp.1957-1960 (Reference 3)) describes the presence and absence of pulses and the polarity (+, −) of the noise vector.
It is a simple structure that can be expressed simply. The algebraic structure codebook has features that it is not necessary to store code vectors and the amount of calculation is small, as compared with a method using a noise codebook storing a plurality of noise vectors. Since the sound quality is comparable to that of the conventional system, it has recently been used in various standard systems.

[0007]

However, in the algebraic structure codebook, as the bit rate of the coding (coding rate) decreases, the deterioration of the sound quality becomes conspicuous. One of the reasons is lack of pulse position information.
In other words, although the position information of the pulse is algebraically simplified in the algebraic structure codebook, the above advantage is obtained. However, at a low coding rate, a position candidate exists at a position where a pulse does not need to be raised, May not exist,
Not only is the efficiency inferior, but also the quality of the voice is degraded.

Another reason why sound quality is degraded when an algebraic structure codebook is used is that the number of pulses is insufficient. When the number of pulses is insufficient, the noise of “bubble wrap” becomes noticeable in the decoded speech. This is because the drive signal is generated from the pulse train, and as the number of pulses decreases, the presence / absence of the pulse becomes more audible. In order to improve the sound quality, it is necessary to reduce the bubble wrap.

As described above, the conventional algebraic structure codebook has the advantages of a simple structure and a small amount of calculation. On the other hand, at a low coding rate, the position information of the pulse train constituting the driving signal of the synthesis filter and There is a problem that the sound quality of the decoded voice is deteriorated due to the shortage of the number of pulses.

An object of the present invention is to provide a speech encoding / decoding method capable of obtaining good sound quality even at a low encoding rate.

[0011]

According to the present invention, there is provided a speech signal comprising a step of generating at least information indicating characteristics of a synthesis filter, and a signal for driving the synthesis filter.
Generating a drive signal including a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates that adaptively change according to the properties of the audio signal. Provide a method of conversion.

According to the present invention, a drive signal including a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates which adaptively change according to the characteristics of an audio signal is synthesized by a filter. And a voice decoding method for decoding a voice signal by inputting the voice signal to a voice signal.

[0013] In the speech encoding / decoding method according to the present invention, the driving signal for driving the synthesis filter has a predetermined number of pulse positions selected from pulse position candidates that adaptively change according to the properties of the audio signal. It includes a pulse train generated by arranging pulses. More specifically, the pulse position candidates are arranged so that there are more candidates as the power of the audio signal increases.

Further, the drive signal includes a pulse train generated by arranging pulses at all the pulse position candidates which adaptively change according to the characteristics of the audio signal, and optimizing the amplitude of each pulse by a predetermined means. Can also be configured. In this case, more specifically, the pulse position candidates are arranged so that there are more candidates as the power of the audio signal is higher.

Further, the drive signal is a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from first pulse position candidates which adaptively change according to the characteristics of the audio signal, or Any of the pulse trains generated by arranging pulses at a predetermined number of pulse positions selected from the second pulse position candidates that are part or all of the positions not used as the first pulse position candidates It can also be generated using. In this case, the first pulse position candidates are more specifically arranged such that there are more candidates as the power of the audio signal increases.

When the drive signal is composed of a pitch vector and a noise vector, the pulses are arranged at a predetermined number of pulse positions selected from pulse position candidates in which the noise vector changes according to the shape of the pitch vector. Generated by In this case, more specifically, the pulse position candidates are arranged such that the greater the power of the pitch vector, the more candidates there are.

Further, a noise vector is generated using a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from position candidates set based on a position candidate density function obtained from a pitch vector shape. It is also possible to adopt a configuration. In this case, more specifically, the pulse position candidates are arranged such that the larger the value of the position candidate density function is, the more candidates exist, and the position candidate density function is the probability that the power of the pitch vector and the pulse are arranged. Is a function obtained in advance.

Further, when a correction means such as a pitch period emphasis filter is used for the noise vector, the pitch vector is selected from pulse position candidates which change according to the shape of the inversely corrected pitch vector obtained by performing processing based on this inverse characteristic. It is generated by arranging pulses at predetermined predetermined number of pulse positions. In this case, more specifically, the pulse position candidate
The arrangement is such that the greater the power of the inverse correction pitch vector, the more candidates there are.

As described above, by changing the pulse position candidates adaptively according to the properties such as the power distribution of the audio signal, the algebraic structure codebook in which the pulse positions and the number of pulses are reduced by the lower coding rate is used. In this case, the coding efficiency is improved, and the coding rate can be reduced while maintaining the sound quality of the decoded speech. In addition, by using a pitch vector to generate a pulse position candidate, it is possible to adapt the pulse position candidate without requiring additional information.

In another speech encoding / decoding method according to the present invention, when the drive signal comprises a pitch vector and a noise vector, the signal is shaped by pulse shaping means having characteristics determined based on the shape of the pitch vector. A drive signal including a pulse train is generated.

With such a configuration, the pulse-like noise included in the decoded speech due to the decrease in the number of pulses is reduced,
Even when the pulse position and the number of pulses are reduced by the lower encoding rate, the lower encoding rate can be achieved while maintaining the sound quality of the decoded speech.

Further, in the speech encoding / decoding method according to the present invention, the pulses are arranged at a predetermined number of pulse positions selected from pulse position candidates which adaptively change according to the characteristics of the speech signal. May be generated, and the pulse train may be shaped by a pulse shaping means having characteristics determined based on the shape of the pitch vector.

[0023]

FIG. 1 shows a speech coding system to which a speech coding method according to a first embodiment is applied. This speech coding system has input terminals 101, 10
6, LPC analysis unit 110, and LPC quantization unit 111
, An LPC synthesis unit 120, an auditory weighting unit 130,
An adaptive codebook 141, a pulse position candidate search unit 142,
It comprises an adaptive algebraic structure codebook 143, a code selector 150, a pitch period enhancer 160, gain multipliers 102 and 103, and adders 104 and 105.

An input speech signal to be encoded is input to an input terminal 101 in units of one frame length, and the LPC analysis unit 110 performs linear prediction analysis in synchronization with the input speech signal, thereby obtaining vocal tract characteristics. Linear prediction coefficient (LPC
Coefficient). The LPC coefficient is calculated by the LPC quantization unit 11
1, the quantized value is input to the LPC synthesizing unit 120 as synthesis filter information indicating the characteristics of the LPC synthesizing unit 120, and an index A indicating the quantized value is used as a coding result by a multiplexing unit (not shown). Output to

The adaptive codebook 141 stores drive signals that have been input to the LPC synthesis section 120 in the past. LP
The drive signal input to the C synthesizing unit 120 is a signal obtained by quantizing a prediction residual signal in the linear prediction analysis, and corresponds to a vocal cord signal including information on a pitch of a sound. Adaptive codebook 1
Reference numeral 41 cuts out a waveform having a length corresponding to the pitch period from the past drive signal, and repeats this to generate a pitch vector. The pitch vector is usually obtained in units of subframes obtained by dividing a frame into several parts.

The pulse position candidate search section 142 calculates, based on the pitch vector obtained by the adaptive codebook 141, at which position in the subframe the pulse position candidate is to be set, and calculates the result in the adaptive algebraic structure. Output to codebook 143.

The adaptive algebraic structure codebook 143 is designed to minimize the distortion of the input speech signal, from which the influence of the pitch vector is subtracted, from the pulse position candidates input from the pulse position candidate search section 142 under the auditory weight. Then, a predetermined number of pulse positions and their signs are searched.

The pulse train output from the adaptive algebraic structure codebook 143 is periodicized in pitch units by the pitch period emphasizing unit 160 as necessary. Pitch cycle emphasis unit 160
Then, the pitch period information L obtained by searching the adaptive codebook 143 is input from the input terminal 106, and the pulse train is given the periodicity of the pitch period.

The pitch vector output from the adaptive codebook 141 and the output from the adaptive algebraic structure codebook 143,
In addition, the pulse train given periodicity by the pitch period emphasizing unit 160 is multiplied by the gain G0 for the pitch vector and the gain G1 for the noise vector by the gain multiplying units 102 and 103, respectively, and then added to the adding unit 104.
And input to the LPC synthesis unit 120 as a drive signal. Note that, as the gains G0 and G1, usually, an optimum gain is selected from a gain codebook (not shown) storing a plurality of gains.

The code selecting section 150 outputs the adaptive codebook 14
1, an index B indicating a pitch vector selected in the search for the index 1, an index C indicating a pulse train selected in the search for the adaptive algebraic codebook 143, and an index indicating the gains G0 and G1 selected in the search for the gain codebook. G is output. Each of these indices B, C,
G and the index A indicating the synthesis filter information, which is the quantization value of the LPC coefficient from the LPC quantization unit 111, are multiplexed by a multiplexing unit (not shown) and output as a bit stream.

Next, the pulse position candidate search unit 142 and the adaptive algebraic structure codebook 143, which are characteristic parts of the present embodiment, will be described.

In the present embodiment, even if the position where the pulse rises at the time of the low coding rate is limited, in order to reduce only the coding rate without deteriorating the sound quality unlike the related art, the pulse is used. The pulse position candidates are set for each sub-frame such that the position where the power of the drive signal is high is concentrated and more position candidates are allocated to the portion where the power of the drive signal is high.

Since the pitch vector resembles the shape of an ideal drive signal, the pulse position candidate search unit 1 based on the pitch vector obtained by searching the adaptive codebook 141.
Setting pulse position candidates at 42 is effective.
Since the same pitch vector is required on the decoding side as on the encoding side, it is not necessary to generate extra additional information with the adaptation of the pulse position candidates.

When adapting the pulse position candidates, if the position candidates are assigned only to the places where the power is large, the sound quality may be degraded because the position candidates do not exist continuously in the section where the power is small. Various methods of adapting the pulse position candidates are conceivable. For example, adaptation with less sound quality degradation is possible by adopting the following method.

With reference to the flowchart shown in FIG. 2, a description will be given of a processing procedure for adaptation of pulse position candidates by the pulse position candidate search unit 142. FIG. 3 shows the input pitch vector waveform (F0), the power (F1) of this input pitch vector waveform, and the smoothed power (F
2) A value (F3) obtained by integrating the smoothed power in the sample direction is shown in FIG.

Similar processing can be performed by using other scales representing the waveform shape such as the absolute value of the amplitude value (the square root of the power) in addition to the power. In the present invention, these are collectively represented by power.

First, the power (F1) is calculated for the input pitch vector (F0) of FIG. 3 (step S).
1) Then, the power (F1) is smoothed, and the smoothed power (F1)
2) is obtained (step S2). For power smoothing, for example, there is a method of taking a moving average by weighting in a window of several samples.

Next, the power smoothed in step S2 is integrated in the sample direction (step S3). This state is shown in (F3) of FIG. Specifically, assuming that the smoothed power of the n-th sample is p (n), the integrated value of the smoothed power p (n) is q (n), and the subframe length is L, the integrated value is q (n) is q (n) = p (n) + q (n-1) + C (n = 0,
.., L-1). Here, C is a constant, and adjusts the degree of bias in the density of the pulse position candidates.

Next, a pulse position candidate is calculated using the integrated value q (n) (step S4). In this case, the integral value is normalized so that the number of position candidates for which the integral value in the final sample is obtained is M. The position of the m-th candidate is shown in FIG.
By associating with the integral value as shown in (F3),
It can be obtained as Sm. By repeating the processing until m = 0,..., M−1, M position candidates can be obtained.

FIG. 4 shows the relationship between the pulse position candidates thus obtained and the power of the pitch vector. The solid line indicates the power envelope of the pitch vector, and the arrows indicate pulse position candidates. As shown in the figure, the distribution of the pulse position candidates becomes dense where the power of the pitch vector is large, and becomes sparse as the power becomes small. As a result, where the power of the pitch vector that is important in sound quality is large, the pulse position can be selected more accurately. Further, even if the number of pulse position candidates decreases due to the lowering of the encoding rate, high-quality encoding becomes possible by adaptively concentrating a small number of pulse position candidates in a place where the power of the pitch vector is large.

Next, the position candidates thus obtained are distributed for each channel (step S5). Although there are various distribution methods, as shown in (F4) of FIG. 3, it is desirable that the position candidates are distributed so that each channel is alternated. Thus, adaptive algebraic structure codebook 143
Is required. In the search, the adaptive algebraic structure codebook 14
The optimum position and code are selected one pulse at a time from each of the three channels (Ch1, Ch2, Ch3), and a noise vector composed of three pulses is generated.

When the subframe length is 80 samples, even if the pulse candidate positions are reduced to about 40 samples in all channels in total, almost no auditory deterioration is perceived by using the above method.

In the algebraic codebook, the pulse amplitude is usually +
Reference 4 (Chang Deyu), which uses either a pulse having amplitude information, which is either 1 or -1.
an, "An 8kb / s low complexity ACELP speech codec,"
1996 3rd International Conference on Signal Proces
sing, pp. 671-4, 1996), there is a method of selecting the pulse amplitude from 1.0, 0.5, 0, -0.5, and -1.0. Reference 5 (K. Oza
wa and T. Araseki, "Low Bit Rate Multi-pulse Speec
h Coder with Natural Speech Quality, "IEEE Proc. I
CASSP'86, pp. 457-460, 1986), a multi-pulse method, which is a kind of pulse sound source, also has a drive signal composed of a pulse train having an amplitude. The present invention is also applicable to a case where a pulse as represented by these examples has an amplitude.

Next, a speech decoding system corresponding to the speech encoding system of FIG. 1 will be described with reference to FIG.

The parts having the same functions as those in FIG. 1 are denoted by the same reference numerals and described. The speech decoding system in FIG.
C combining section 120, LPC inverse quantization section 121, adaptive codebook 141, pulse position candidate searching section 142, adaptive algebraic structure codebook 143, pitch period emphasizing section 160, gain multiplying sections 102 and 103, and The coded stream, which is constituted by the adder 104 and transmitted from the speech coding system of FIG. 1, is input.

The input coded stream is input to a demultiplexing unit 121 (not shown). The demultiplexing unit 121 calculates the index A of the synthesis filter information and the pitch vector selected in the search for the adaptive codebook 141. Index B, index C representing the pulse train selected in the search for adaptive algebraic structure codebook 143, index G indicating the gains G0 and G1 selected in the search for gain codebook, and index L indicating the pitch period.
It is separated and taken out.

The index A is the LPC inverse quantizer 12
The LPC coefficient, which is decoded by 1 and is synthesis filter information, is obtained and input to the LPC synthesis unit 120. Indexes B and C are input to adaptive codebook 141 and adaptive algebraic structure codebook 143, respectively.
A pitch vector and a pulse train are output from 1,143. In this case, the adaptive algebraic structure codebook 143 obtains a pulse position and a code from the adaptive algebraic structure codebook 143 and the index B generated by the pulse position candidate search unit 142 based on the pitch vector input from the adaptive codebook 141. And outputs a pulse train. Adaptive Algebraic Codebook 1
The pulse train output from 43 is given a periodicity of the pitch period L by the pitch period emphasizing unit 160 as necessary.

The pitch vector output from the adaptive codebook 141 and the output from the adaptive algebraic structure codebook 143,
In addition, the pulse train given periodicity by the pitch period emphasizing unit 160 is multiplied by the gain G0 for the pitch vector and the gain G1 for the noise vector by the gain multiplying units 102 and 103, respectively, and then added to the adding unit 104.
Are input to the LPC synthesis section 120 as a drive signal, and the LPC synthesis section 120 outputs a reproduced audio signal. The gains G0 and G1 are selected from a gain codebook (not shown) according to the index G.

As described above, according to the present embodiment, it is possible to reduce only the bit rate while maintaining the voice quality, and realize high-quality voice coding / decoding at a low coding rate. Can be.

FIG. 6 shows a speech coding system according to a second embodiment of the present invention. This speech coding system is different from the configuration shown in FIG. 1 according to the first embodiment in that the pulse position candidate search unit 142 and the adaptive algebraic structure codebook 143 are used.
, A general noise codebook 144 is provided as an alternative to the adaptive algebraic structure codebook 143, and a pulse shaping filter analyzing unit 161 and a pulse shaping unit 162 are added.

Next, the processing procedure of the present embodiment will be described. After performing LPC analysis and LPC quantization of the input speech signal, the process up to the point of searching the adaptive codebook 141 is as follows.
This is the same as the first embodiment. In this example, the noise codebook 144 is constituted by, for example, an algebraic structure codebook.

The pulse shaping filter analysis unit 161 determines and outputs the filter coefficient of the pulse shaping unit 162 based on the pitch vector obtained by searching the adaptive codebook 141. The pulse shaper 162 shapes the output of the noise codebook 144 and outputs the result as a noise vector.

As in the first embodiment, if necessary, the noise vector is periodicized by using the pitch period emphasizing unit 160, and the gains G0, G0,
G1 is determined and an index is output. Since the filter coefficient of the pulse shaping section 162 is obtained from the pitch vector, no new additional information is required.

The feature of this embodiment is that the pulse shaping section 162
Is set based on the pitch vector waveform, and pulse shaping is performed on the pulse train output from the noise codebook 144 including the algebraic structure codebook. As described in the first embodiment, the pulse position and the number of pulses are reduced as the encoding rate is reduced, and the deterioration of the sound quality becomes conspicuous. When the number of pulses is reduced, the noise of “bubbles” becomes noticeable in the decoded speech, but the pulse shaping section 162 as in the present embodiment.
By using this, the bubble wrap feeling is greatly reduced.

The design method of the pulse shaping unit 162 is as follows.
Various methods can be used. As a first example, a method utilizing the property that when a drive signal for driving a synthesis filter is phase-equalized, the signal becomes a pulse-like signal is considered. When an inverse filter for phase equalization is used, a pulse-like signal is input to obtain a drive signal-like waveform. The disadvantage of using the conventional pulse waveform is that the phase information included in the ideal drive signal is lacking. This problem becomes conspicuous as the number of pulses decreases. Therefore, by adding the phase information by the pulse shaping unit 162 as in this example, a waveform closer to an ideal driving signal can be generated from the pulse waveform.

In this first example, it is necessary to transmit the information of the filter coefficient of the phase equalization inverse filter, and the coding rate (bit rate) is increased by that amount. Therefore, as a second example of the pulse shaping unit 162, a method using a pitch vector as an approximation of the phase information is considered. Since the pitch vector has a similar shape to the drive signal in a sound section or the like, phase information can be extracted.

As a specific method, a pulse shaping filter that determines a synchronization point such as a peak position of a pitch vector, extracts a waveform of several samples from the synchronization point, and uses the waveform as an impulse response can be used. . The effect appears when the length of the extracted waveform is about 2 to 3 samples.
It is also effective to apply a window to the sample taken out and attenuate it. Further, since the same pitch vector can be obtained on the decoding side as on the encoding side, there is also an advantage that a new transmission bit is not required. Noise codebook 14
Since the pulse shaping section 162 is constant during the search of 4, the amount of calculation can be reduced by calculating the impulse response in advance together with the LPC synthesis section 120.

FIG. 7 shows a speech decoding system corresponding to the speech encoding system of FIG. The parts having the same functions as those in FIG. 6 will be described with the same reference numerals. The speech decoding system in FIG. 7 includes an LPC synthesis unit 120, an LPC inverse quantization unit 121, an adaptive codebook 141, an algebraic structure code , A pulse shaping filter analyzing unit 161, a pulse shaping unit 162, a pitch period emphasizing unit 160, gain multiplying units 102 and 103, and an adding unit 10.
4, and receives an encoded stream transmitted from the speech encoding system of FIG.

The input coded stream is input to a demultiplexer (not shown), and the demultiplexer outputs the index A of the synthesis filter information and the adaptive codebook 1 described above.
An index B indicating a pitch vector selected in the search for 41, an index C indicating a pulse train selected in the search for the noise codebook 144, and an index G indicating the gains G0 and G1 selected in the search for the gain codebook are separated. It is taken out. The pitch period L is calculated from the index B.

The index A is the LPC inverse quantizer 12
1 and becomes synthesis filter information, which is input to the LPC synthesis section 120. The indexes B and C are input to the adaptive codebook 141 and the noise codebook 144, respectively, and a pitch vector and a pulse train are output from these codebooks 141 and 144.

In this case, the pulse train output from the random codebook 144 is converted to a pulse shaping filter analysis unit 16 based on the pitch vector obtained by searching the adaptive codebook 141.
After being processed by the pulse shaping unit 162 whose coefficient is set to 1, the pitch period emphasizing unit 160 gives periodicity of the pitch period L as necessary.

The pulse train output from the adaptive codebook 141 and the pulse train output from the noise codebook 144 and passed through the pulse shaping unit 162 and the pitch period emphasizing unit 160 are subjected to gain multiplications 102 and 103 to obtain gains G0 and G0 for the pitch vector. Gain G1 for noise vector
Are added to each other by the adder 104, input to the LPC synthesizer 120 as a drive signal, and the LPC synthesizer 120 outputs a decoded speech signal synthesized. The gains G0 and G1 are selected from a gain codebook (not shown) according to the index G.

As described above, according to the present embodiment, the use of the pulse shaping section 162 enables the decoding of the decoded speech even when the algebraic structure codebook in which the number of pulses is reduced due to the lower coding rate is used as the noise codebook 144. , It is possible to effectively reduce only the coding rate while maintaining the sound quality.

FIG. 8 shows a speech coding system according to a third embodiment of the present invention. This speech coding system includes a pulse shaping filter analyzing unit 161 and a pulse shaping unit 162 described in the second embodiment in the configuration of the first embodiment.
Has been added.

Next, the processing procedure of this embodiment will be described. As in the first embodiment, first, LPC analysis and LPC quantization are performed, and after the search of the adaptive codebook 141 is completed, the pitch vector becomes pulsed. Position candidate search unit 142
Is passed to the pulse shaping filter analyzer 161. The pulse position candidate search unit 142 obtains pulse position candidates using the method described in the first embodiment, and creates an adaptive algebraic structure codebook 143. Pulse shaping filter analyzer 16
1, the pulse shaping unit 1 as described in the second embodiment.
A coefficient of 62 is determined.

In the search for adaptive algebraic structure codebook 143, the output pulse train is shaped by pulse shaping section 162.
In an actual search, the impulse responses of the pulse shaping unit 162 and the pitch period emphasizing unit 160 are combined with the LPC synthesizing unit 120 to reduce the amount of calculation.

FIG. 9 shows a speech decoding system corresponding to the speech encoding system of FIG. Since the operation of the speech decoding system is obvious from the operation of the speech decoding system described in the first and second embodiments, the same reference numerals are given to the same parts as in FIGS. Detailed description is omitted.

As described above, in the present embodiment, the pulse position candidate searching unit 142 and the adaptive algebraic structure codebook 143 described in the first embodiment, and the pulse shaping filter analyzing unit 161 and the pulse Shaping unit 162
Are used at the same time, it is possible to maintain high sound quality even when a small number of pulses are set at limited position candidates, and it is possible to realize a voice coding method with high sound quality and a low coding rate.

FIG. 10 is a block diagram showing a speech coding system according to the fourth embodiment of the present invention. In this speech coding system, the pulse position candidate search unit of the first embodiment is configured by a pitch vector smoothing unit 171, a position candidate density function calculation unit 172, and a position candidate calculation unit 173, except that It has the same configuration as the embodiment.

Next, the processing procedure of this embodiment will be described. As in the first embodiment, after the LPC analysis and LPC quantization and the search of the adaptive codebook 141 are completed, the pitch vector is changed to the pulse position. The pitch vector is passed to the pitch vector smoothing unit 171 of the candidate searching unit 142. The pitch vector smoothing unit 171 performs, for example, the processing of steps S1 and S2 in the flowchart of FIG. 2 on the pitch vector, obtains the power envelope of the pitch vector, and outputs this. The position candidate density function calculation unit 172 converts the power envelope into a position candidate density function and outputs it. The position candidate calculation unit 173 calculates a pulse position candidate using the position candidate density function instead of the power envelope, and creates an adaptive algebraic structure codebook 143 according to the obtained pulse position candidate. Subsequent processing is the same as in the first embodiment.

The feature of the present embodiment lies in the processing method of the pulse position candidate search section 142. In the first embodiment, the adaptation of the pulse position candidates is performed using the power envelope of the pitch vector as it is. In the present embodiment, the power envelope is converted into a position candidate density function, and then the adaptation is performed using the function. Is going. This will be described in detail with reference to FIG. FIG.
(A) is the power envelope of the pitch vector output from the pitch vector smoothing unit 171. In the position candidate density function calculating unit 172, the power envelope of the pitch vector (FIG. 11)
A position candidate density function (FIG. 11B) is generated from (a)). At this time, the conversion is performed using a function f indicating the correspondence between the value (x) of the power envelope and the value (f (x)) of the position candidate density function shown in FIG. As a method of creating the function f, for example, there is a method of statistically obtaining a number by processing many learning voices. Further, table data or the like can be used instead of the function.

The pulse position candidate search section 142 prepares the same for the encoder and the decoder together with the conversion function f, so that there is no need to send information on adaptation.
There is no increase in the bit rate as compared to the case without adaptation.

FIG. 12 shows the configuration of the speech decoding system of the present embodiment corresponding to the speech encoding system of FIG.
Since the operation of the speech decoding system is obvious from the operation of the speech decoding system described in the first to third embodiments, detailed description will be omitted.

As described above, in the present embodiment, since the value of the power envelope of the pitch vector and the density of the pulse position candidates are converted using the function f, the processing procedure is slightly complicated as compared with the first embodiment. Thus, more accurate distribution of position candidates is possible. In the first embodiment, x
= F (x).

FIG. 13 is a block diagram showing a speech coding system according to the fifth embodiment of the present invention. In this speech coding system, except that the pulse position candidate search unit of the first embodiment includes a pitch filter inverse operation unit 174, a smoothing unit 175, and a position candidate calculation unit 173,
This is the same configuration as the embodiment.

Next, the processing request of the present embodiment will be described. First, as in the first embodiment, after the LPC analysis and LPC quantization and the search of the adaptive codebook 141 are completed, the pitch vector becomes pulsed. It is passed to the pitch filter inverse operation unit 174 of the position candidate search unit 142. The pitch filter inverse operation unit 174 performs an operation representing the inverse characteristic of the pitch cycle emphasis unit 160. For example, the transfer function P of the pitch filter
When (Z) is given by P (z) = 1−az ＾ (− L) (1), the pitch filter inverse operation unit 174 sets the transfer function Q (z) to Q (Z) = 1 / (1-baz). ＾ (− L)) (2) A method using the filter given by Here, a is a constant, b is the degree of the inverse characteristic, and when b = 1, Q
(Z) is an inverse filter of P (z). The input pitch vector is output after being subjected to an inverse operation, and the power envelope is obtained by the smoothing unit 175 in the same manner as the pitch vector smoothing unit 171 of the fourth embodiment. Position candidate calculation unit 1
At 73, a pulse position candidate is selected according to the power envelope, and an adaptive algebraic structure codebook 143 is created. Subsequent processing is the same as in the first embodiment.

The feature of this embodiment is that the pitch period emphasizing unit 16
The point is that the pitch vector considering the influence of 0 is used for adapting the pulse position candidate. The reason why the efficiency is improved by doing this will be described.

The noise vector generated from the adaptive algebraic structure codebook is pitch-performed by the pitch period emphasizing unit 160. When the equation (1) is used for the periodization, the pulse near the head of the subframe is repeated many times within the subframe at the pitch cycle interval, whereas the number of repetitions decreases in the latter half of the pulse. By observing the actually obtained noise code vector, it can be confirmed that the pulse tends to be more likely to appear near the head as the strong pitch filter is used. This indicates that the pulse position is deeply related not only to the pitch vector shape but also to the pitch filter. In the present embodiment, the pitch cycle emphasis unit 160 is used by using the pitch filter inverse operation unit 174.
Adaptation of the pulse position candidates taking into account the influence of is realized.

In the third embodiment, it is possible to apply two types of filters, a pulse shaping filter and a pitch filter, to the noise vector. When the present embodiment is applied to such a case, it is ideal that a characteristic obtained by combining two filters is obtained, and an inverse characteristic of the characteristic is used for the pitch filter inverse operation unit. However, the effect can be obtained only by using the characteristic of the pitch filter which has a large influence because the processing amount increases. Further, the order of the pitch filter inverse operation unit 174 and the smoothing unit 175 can be reversed.

FIG. 14 shows the configuration of the speech decoding system of the present embodiment corresponding to the speech encoding system of FIG.
The operation of this speech encoding system is obvious from the operation of the speech decoding system described in the first to fourth embodiments, and therefore detailed description is omitted.

FIG. 15 is a block diagram showing a speech coding system according to the sixth embodiment of the present invention. This speech coding system has the same configuration as that of the first embodiment, except that the adaptive algebraic structure codebook of the first embodiment is replaced by a noise vector generator 180 and an amplitude codebook 181.

Next, the processing procedure of this embodiment will be described. As in the first embodiment, after the LPC analysis and LPC quantization and the search of the adaptive codebook 141 are completed, the pitch vector is changed to the pulse position search. It is passed to the unit 174. The pulse position search unit 174 obtains a pulse position based on the power envelope of the pitch vector in the same manner as in the first embodiment, and outputs this to the noise vector generation unit. Here, the present embodiment is different from the previous embodiments in that the noise vector search unit generates all the pulses at the position obtained by the pulse position search unit 174. That is, in the embodiments described above, pulse position candidates are obtained, and the optimal pulse position is selected from the candidate pulse positions in the adaptive algebraic structure codebook. On the other hand, in the present embodiment, all the pulse position candidates are used simultaneously. . Therefore, the process of selecting the pulse position becomes unnecessary. Instead, the amplitude of each pulse is
Processing to select from 1 is added. Also, the output signal is information D representing the pulse amplitude instead of the information c representing the pulse position.
Is output.

A method of generating a noise vector will be described in detail with reference to FIG. FIG. 16A shows an amplitude pattern obtained from the amplitude codebook by an arrow. In this case, it is assumed that seven pulses are made. 16 (b) and FIG.
The waveform of (c) is the pitch vector power envelope obtained by the pulse position search unit 174 and the corresponding pulse position (indicated by a circle in the figure). In FIG. 16 (b), there are two peaks in the power, so that seven pulse positions are dispersed in two places, whereas in FIG. 16 (c), there is one peak in the center, so that the pulse position is in the center. focusing. FIG. 16D and FIG.
(E) is a noise vector in which a pulse having the amplitude of FIG. 16 (a) is set at each pulse position. It can be seen that the shape of the drive signal also changes according to the pitch vector power envelope. As described above, since the power envelope information of the pitch vector does not need to be transmitted, in the present embodiment, the shape of the noise vector can be approximated to the ideal noise vector shape without increasing the bit rate.

In this embodiment, as the bit rate increases, more pulse amplitude information D can be sent and the quality is improved, but the degree of improvement is reduced. At a somewhat higher bit rate, performance may be improved by including noise vectors that have pulses at unselected positions rather than increasing the amplitude information as search candidates. Specifically, the pulse position search unit 174 outputs a pattern (pulse pattern) of a different pulse position, and the noise vector generation unit searches for an amplitude for each pulse pattern. As the pulse pattern, in addition to the pulse pattern adapted to the above-described pitch vector, a pulse pattern generated from a pulse position not selected as the pulse pattern is prepared. For example, there is a method in which the remainder obtained by subtracting the sample position selected by the adaptation from all the sample positions of the sub-frame is used as a second pulse pattern to search the amplitude of two types of pulse patterns. The number of bits assigned to the amplitude information can be different for each pulse pattern, and it is generally more efficient to allocate more bits to the pulse pattern using the adaptation. When a plurality of pulse patterns are used, it is necessary to transmit information indicating which pulse pattern is used in the information D, and the amplitude information is reduced accordingly, but only a single pulse pattern is searched. Better quality.

FIG. 17 shows the configuration of the speech decoding system of this embodiment corresponding to the speech encoding system of FIG.
Since the operation of the audio decoding system is obvious from the operation of the audio decoding system described in the first to fifth embodiments, detailed description will be omitted.

Although the speech encoding / decoding method has been described in the above embodiment, the present invention can also be applied to a speech synthesis method, in which case the speech decoding method shown in FIGS. 5, 7 and 9 is used. In the system, each index may be given based on a reproduced audio signal to be synthesized.

[0087]

As described above, according to the present invention, even if an algebraic structure codebook in which the pulse positions and the number of pulses are reduced by reducing the coding rate is used, high-quality speech coding / coding can be performed.
Decryption can be performed.

[Brief description of the drawings]

FIG. 1 is a block diagram of a speech encoding system according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing a procedure for selecting a pulse position candidate according to the first embodiment;

FIG. 3 is a view showing a state of processing in each step of FIG. 2;

FIG. 4 is a diagram showing a relationship between a power envelope of a pitch vector and a pulse position candidate in the first embodiment.

FIG. 5 is a block diagram of a speech decoding system according to the first embodiment;

FIG. 6 is a block diagram of a speech encoding system according to a second embodiment of the present invention.

FIG. 7 is a block diagram of a speech decoding system according to a second embodiment;

FIG. 8 is a block diagram of a speech encoding system according to a third embodiment of the present invention.

FIG. 9 is a block diagram of a speech decoding system according to a third embodiment;

FIG. 10 is a block diagram of a speech coding system according to a fourth embodiment of the present invention.

FIG. 11 is a diagram showing the relationship between the values of the pitch vector power envelope, the position candidate density function, the power envelope, and the position candidate density function.

FIG. 12 is a block diagram of a decoding system according to a fourth embodiment;

FIG. 13 is a block diagram of a speech coding system according to a fifth embodiment of the present invention.

FIG. 14 is a block diagram of a decoding system according to a fifth embodiment;

FIG. 15 is a block diagram of a speech coding system according to a sixth embodiment of the present invention.

FIG. 16 is a diagram for explaining a noise vector generation method.

FIG. 17 is a block diagram of a decoding system according to a sixth embodiment.

[Explanation of symbols]

 101 voice input terminal 102, 103 gain multiplying unit 104, 105 adding unit 110 LPC analyzing unit 111 LPC quantizing unit 120 LPC synthesizing unit 130 auditory weighting unit 141 adaptive codebook 142 pulse position candidate search Unit 143: adaptive algebraic structure codebook 144: noise codebook 150: code selection unit 160: pitch period emphasis unit 161: pulse shaping filter analysis unit 162: pulse shaping unit 171 ... pitch vector smoothing unit 172: position candidate density function calculation unit 173: position candidate calculating unit 174: pulse position searching unit 180: noise vector generating unit 181: amplitude codebook

Claims (20)

[Claims] 1. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information indicating characteristics of a synthesis filter and a drive signal for driving the synthesis filter, wherein the drive signal is A speech encoding method comprising a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates that adaptively change according to properties. 2. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information representing characteristics of a synthesis filter and a drive signal for driving the synthesis filter, wherein the drive signal is It is characterized by comprising a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates arranged such that there are more candidates as the power is larger. The audio coding method to use. 3. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information representing characteristics of a synthesis filter and a drive signal for driving the synthesis filter, wherein the drive signal is A pulse is arranged at a predetermined number of pulse positions selected from pulse position candidates that adaptively change according to the property, and the amplitude of each pulse includes a pulse train generated by optimizing by a predetermined means. A speech coding method characterized by being performed. 4. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information indicating characteristics of a synthesis filter and a drive signal for driving the synthesis filter, wherein the drive signal is A pulse train generated by arranging pulses at a predetermined number of pulse positions selected from first pulse position candidates that adaptively change according to properties, or a pulse train generated by the first pulse position candidate was not held. A speech comprising one of pulse trains generated by arranging pulses at a predetermined number of pulse positions selected from second pulse position candidates including part or all of positions. Encoding method. 5. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information indicating characteristics of a synthesis filter and a drive signal including a pitch vector and a noise vector for driving the synthesis filter. The speech code is characterized in that the vector is configured using a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates that change according to the shape of the pitch vector. Method. 6. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information indicating characteristics of a synthesis filter and a drive signal including a pitch vector and a noise vector for driving the synthesis filter. The vector uses a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates arranged such that there are more candidates as the power of the pitch vector increases. A speech coding method characterized by comprising: 7. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information representing the characteristics of a synthesis filter and a drive signal including a pitch vector and a noise vector for driving the synthesis filter. The vector calculates a position candidate density function from the shape of the pitch vector, and generates a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from position candidates set based on the position candidate density function. A speech encoding method characterized by comprising: 8. A speech signal which encodes a speech signal by expressing at least information representing characteristics of a synthesis filter and a drive signal for driving the synthesis filter by a pitch vector and a noise vector whose shape is processed by a correction means. In the encoding method, the noise vector is a predetermined number of pulse positions selected from pulse position candidates that change according to the shape of an inverse correction pitch vector obtained by performing an operation based on the inverse characteristic of the correction means on the pitch vector. A speech coding method characterized by comprising a pulse train generated by arranging a pulse in a speech signal. 9. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information representing characteristics of a synthesis filter and a drive signal including a pitch vector and a noise vector for driving the synthesis filter. A speech encoding method, wherein the signal is configured to include a pulse train shaped by a pulse shaping means having characteristics determined based on the shape of the pitch vector. 10. A speech encoding method for encoding a speech signal by expressing the speech signal with at least information representing characteristics of a synthesis filter and a drive signal comprising a pitch vector and a noise vector for driving the synthesis filter. The vector is generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates arranged so that there are more candidates as the power of the pitch vector increases, and the pitch vector Characterized by comprising a pulse train shaped by a pulse shaping means having characteristics determined based on the shape of the speech. 11. A driving signal including a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates that adaptively change according to the characteristics of an audio signal. A speech decoding method characterized in that a speech signal is decoded by inputting to a synthesis filter. 12. A pulse train generated by arranging pulses at a predetermined number of pulse positions selected from pulse position candidates arranged such that there are more candidates as the power of the audio signal increases. A speech decoding method characterized by inputting a drive signal composed of the following to a synthesis filter to decode a speech signal. 13. A pulse train including a pulse train generated by arranging a pulse having a given amplitude at a predetermined number of pulse positions selected from pulse position candidates that adaptively change according to the characteristics of an audio signal. A speech decoding method characterized by inputting a driving signal to a synthesis filter to decode a speech signal. 14. A pulse train generated by arranging pulses at a predetermined number of pulse positions selected from first pulse position candidates that adaptively change according to the characteristics of an audio signal, or said first pulse. Drive including a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from second pulse position candidates including a part or all of positions not used as position candidates A speech decoding method comprising: inputting a signal to a synthesis filter to decode a speech signal. 15. A noise vector constituted by using a pulse train generated by arranging pulses at a predetermined number of pulse positions selected from a pitch vector and pulse position candidates that change according to the shape of the pitch vector. A speech decoding method characterized by inputting a drive signal consisting of the following to a synthesis filter to decode a speech signal. 16. A pulse vector is generated by arranging pulses at a predetermined number of pulse positions selected from a pitch vector and pulse position candidates arranged such that there are more candidates as the power of the pitch vector is larger. A speech decoding method characterized in that a speech signal is decoded by inputting a drive signal composed of a noise vector formed by using a pulse train to a synthesis filter. 17. A position candidate density function is obtained from a pitch vector and the shape of the pitch vector, and pulses are arranged at a predetermined number of pulse positions selected from position candidates set based on the position candidate density function. A speech decoding method characterized by inputting a drive signal composed of a noise vector formed using a generated pulse train to a synthesis filter to decode a speech signal. 18. A pulse position which is processed by a pitch vector and a correction means, and which is selected from pulse position candidates which change according to the shape of an inverse correction pitch vector obtained by performing an operation based on the inverse characteristic of the correction means on the pitch vector. A speech decoding method characterized by inputting a driving signal composed of a noise vector formed using a pulse train generated by arranging pulses at a predetermined number of pulse positions to a synthesis filter and decoding a speech signal. . 19. A driving signal comprising a pitch vector and a noise vector and including a pulse train shaped by pulse shaping means having characteristics determined based on the shape of the pitch vector is input to a synthesis filter. An audio decoding method comprising decoding an audio signal. 20. A pulse vector generated by arranging pulses at a predetermined number of pulse positions selected from pitch position candidates arranged such that there are more candidates as the pitch vector and the power of the pitch vector are larger. Decoding a voice signal by inputting a drive signal including a noise vector including a pulse train shaped by a pulse shaping unit having characteristics determined based on the shape of the pitch vector to a synthesis filter. A speech decoding method characterized by the following.