JPH09172413A - Variable rate voice coding system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a speech quality without receiving the influence of transmission rate by transmitting coding information of all the class when the transmission rate is the highest and transmitting coding information by canceling coding information of the class of small timewise fluctuation along with the lowering of the transmission rate. SOLUTION: A variable rate transmission frame buffer 2 temporarily stores coding voice information c4 and outputs this to the side of an adaptive modulation means at a rate based on a transmission rate control signal b4. A variable rate voice decoder 4 decode-processes coding voice information g4 to output a reproduced voice signal h4. At this time, coding voice information of the class of small timewise fluctuation, which is not transmitted in a frame showing that a transmission rate control signal f4 is the frame of low rate transmission, is interpolated by information of a frame transmitted before/after it, otherwise after reproducing information of a frame just before as it is, and is voice-decode- processed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to digital mobile communication used in multimedia application systems,
In particular, the present invention relates to a variable rate speech coding method for application to a digital mobile communication method using an adaptive modulation method in which transmission quality is improved by adapting the modulation method according to changes in traffic volume and propagation path conditions. .

[0002]

2. Description of the Related Art Conventionally, in the design of a mobile communication system using a radio transmission line in which the reception level fluctuates significantly due to fading, the modulation method to be applied is determined by the average characteristics of the transmission line. Therefore, when the state of the transmission path is good, the transmission becomes less than the original transmittable capacity and the efficiency becomes poor. Further, when the state of the transmission path is bad, the error rate becomes equal to or higher than the design value, and the quality of reproduced voice deteriorates. In order to solve this problem, the objective was to realize the high throughput transmission with high transmission quality by switching and selecting the optimal modulation method and transmission rate within the allocated band according to the instantaneous transmission path fluctuation. Adaptive modulation schemes have been proposed (Shinya Otsuki et al .: "Transmission characteristics analysis of adaptive modulation schemes using QAM")
See IEICE Technical Report, RCS94-66 (19994-09)). Since this adaptive modulation system is a system that changes the transmission rate according to the state of the propagation path, it is possible to adapt the coding rate of the voice corresponding to the switching of the transmission rate, and Equipment is required.

A variable-rate ADPCM (40/32/24/16) is the main conventional variable-rate speech coding system.
kbps) (ITU-T Recommendation G.726), embedded ADPCM (40/32/24/16 / kbps) (ITU
-T Recommendation G. 727) and QCELP (8/4) proposed by Qualcom in the North American IS-95 system.
/2/0.8kbps) etc.

The above variable rate ADPCM has a voice coding rate of 40/32/24 / in arbitrary sample units.
This is a method of selecting from 16 kbps. In addition, the embedded ADPCM selects a voice coding rate from 40/32/24/16 kbps in arbitrary sample units,
The coded audio information obtained as a result is core information (information indispensable for generating the reproduced sound) and enhancement information (information for improving the quality of the reproduced sound. Even if this information is discarded, the reproduced sound is not reproduced. It can be generated), and all or part of the enhancement information is discarded as necessary, and the reproduced voice can be obtained only from the core information and the remaining enhancement information. Furthermore, QCE
The LP classifies the input speech signal acoustically for each frame (classifies voiced / unvoiced / transient / noise), and determines the optimal speech coding rate for each acoustic property as 8/4/2. /
This is a method of performing a coding process by selecting from 0.8 kbps, and an average coding rate is about 4.8 kbps.

[0005]

Problems to be solved when these conventional variable rate speech coding systems are applied to a communication system based on an adaptive modulation system will be described below. In the adaptive modulation method, the quality of the propagation path is checked by monitoring the reception level and the like, and the transmission rate or the modulation multilevel number is changed based on the result, so the transmission capacity changes momentarily. Along with this, the coding rate of the speech coder is also
The higher the propagation path condition, the higher the value, and the poorer the value, the lower the value. The problem here is that in the above-mentioned conventional coding method, the quality of the reproduced voice also changes when the coding rate of the voice coder changes, so the call quality changes from moment to moment, and The point is that they feel uncomfortable. Further, if the state of the propagation path is further deteriorated, information transmission becomes impossible, and the reproduced voice may be momentarily interrupted. At present, when applied to a system using an adaptive modulation method, there is no variable rate speech coding method in which speech quality is not easily affected by a change in transmission rate.

In order to solve this problem, the object of the present invention can be applied to a communication system using an adaptive modulation system in which the transmission rate is changed in response to changes in the propagation path state, and
It is an object of the present invention to provide a variable rate speech coding system in which speech quality is hardly affected by a change in transmission rate of an adaptive modulation system.

[0007]

According to the variable rate speech coding system of the present invention, a received signal is monitored to detect a state of a propagation path in frame units, and a high transmission rate is designated when the condition of the propagation path is good. Then, a propagation path state estimating means for outputting a transmission rate control signal for decreasing the transmission rate to N stages as the state of the propagation path deteriorates, and an adaptive modulation for changing the transmission rate to N stages by the transmission rate control signal. A variable-rate voice encoding system for application to a communication system of an adaptive modulation system provided with a means and an adaptive demodulating means, wherein a transmitting side encodes an input voice signal by performing a voice encoding process for each frame. After extracting the voice information, the coded voice information is classified into N stages from information having a large temporal variation to information having a small variation in correspondence with the transmission rate control signal. For the frame in which the control signal indicates the highest rate, the encoded voice information of all classes is output to the adaptive modulation means side at the rate indicated by the transmission rate control signal, and the rate indicated by the transmission rate control signal is As the audio data becomes lower, the coded audio information is discarded from the class having the smaller temporal fluctuation according to the degree, and the transmission rate control signal indicates only the remaining coded audio information of the class having the large temporal fluctuation. A variable rate audio encoding means for outputting to the adaptive modulating means side at a rate that the receiving side receives the encoded audio information from the adaptive demodulating means, and based on the transmission rate control signal, in a frame of high rate transmission. Reproduces all classes of coded audio information and has not been transmitted in low-rate transmission frames. After being interpolated by the transmitted information of the frames before and after the frame, or after being reproduced by using the information of the transmitted frame of the immediately preceding frame as it is, transmitted encoded audio information and the reproduced encoding It is characterized by comprising variable rate audio decoding means for generating a reproduced audio signal by performing audio decoding processing of audio information.

Further, in order to improve resistance to transmission errors, the above-mentioned variable rate audio encoding system is characterized by adding an error correction / detection function to the encoded audio information.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION

(Operation) In the present invention, on the transmission side, the input speech signal is coded for each frame, and the resulting coded speech information is classified into classes according to the magnitude of temporal fluctuation. When the propagation path state is good, all classes are classified. Is output to the modulation means side, the coded voice information is discarded from the class with the smaller temporal fluctuation as the propagation path condition worsens, and only the information of the remaining class with large temporal fluctuation is applied. The receiving side receives the coded voice information through the adaptive demodulating means, and the receiving side receives the coded voice information through the adaptive demodulating means, and the coded voice of a class with small temporal fluctuation that was not transmitted in a frame with a bad propagation path state. For information, it is interpolated by the information of the frames transmitted before and after it or reproduced using the information of the frame transmitted immediately before, and then the encoded audio information transmitted. Generates audio decoding processing reproduced audio signal by using the finely the reproduced coded speech information. As a result, it is possible to reduce the drop in the reproduced voice quality when the state of the propagation path is poor, and it is possible to realize speech quality that is not easily affected by the state of the propagation path. Further, by adding a function of performing error correction / detection at a variable rate to the encoded voice information according to the transmission rate, it is possible to improve resistance to transmission errors.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment of the variable rate speech coding system and its apparatus of the present invention will be described with reference to FIGS. FIG. 1 is a configuration diagram showing a first embodiment of the present invention described in claim 1. FIG. 2 is a block diagram showing a second embodiment of the present invention as set forth in claim 6, in order to improve resistance to a transmission error occurring in mobile radio communication, a channel coding / decoding function (error correction. A configuration including a detection function) is shown. 3 and 4 are detailed specific examples of the speech encoder and the speech decoder shown in FIGS. 1 and 2, which are coded linear prediction (CELP: Codec Excited Linear Prediction).
It is a structural example figure at the time of using (claim 3). further,
The invention according to claim 7 is applied to the first and second embodiments of the present invention shown in FIGS. 1 and 2, and the modulation multilevel number is changed to N stages for each frame in accordance with the propagation path condition. The present invention is applied to an adaptive modulation system that changes the transmission rate in N steps.

First, a first embodiment will be described. FIG. 1 is a diagram showing an example of the configuration of a variable rate speech coding system and its apparatus described in claim 1 of the present invention. The figure (A) shows a variable rate speech coding apparatus used on the transmitting side, and the figure (B) shows a decoding apparatus on the receiving side. In the figure, 1 is a variable rate speech encoder on the transmitting side, 2 is a variable rate transmitting frame buffer, 3 is a variable rate receiving frame buffer on the receiving side, and 4 is a variable rate speech decoder.

The variable rate speech coder 1 of the transmission side coding apparatus of FIG. 1A is sampled at 8 kHz and
The input voice signal a4 quantized with a precision of 2 bits or more is encoded, and the resulting encoded voice information is classified into classes according to the magnitude of temporal fluctuation, and the transmission rate control signal b4
The encoded audio information c4 of the class selected based on the above is output.

Here, the transmission rate control signal b4 is a channel state estimation means (one component of the adaptive modulation system) for estimating the state of the channel based on the reception level and the like on the receiving side and determining the optimum transmission capacity. Is a control signal output from
Predetermined unit time (corresponding to frame length: eg 5 mse
It is a signal that specifies the transmission rate to the variable rate speech encoding device and the decoding device of the present invention for each c). The detailed configuration of the variable rate speech encoder 1 will be described later.

The variable rate transmission frame buffer 2 temporarily stores the coded voice information c4 and outputs the coded voice information d4 to the adaptive modulation means side at a rate based on the transmission rate control signal b4.

The variable rate reception frame buffer 3 of the variable rate audio decoding device used on the receiving side in FIG.
Is coded voice information e4 that is transmitted from the variable rate voice coding device on the transmission side and is error-protected, is input from the adaptive demodulation means side, is temporarily stored, and is coded based on the rate specified by the transmission rate control signal f4. The converted voice information g4 is output. Here, the transmission rate control signal f4 is the transmission rate control signal b referred to by the above-mentioned variable rate speech coder.
Same as 4. The variable rate audio decoder 4 decodes the encoded audio information g4 and outputs a reproduced audio signal h4. At this time, the encoded voice information of the class with small temporal fluctuation that was not transmitted in the frame indicating that the transmission rate control signal f4 is a low-rate transmission frame was transmitted before and after that. After being interpolated by frame information or reproduced using the information of the frame transmitted immediately before, audio decoding processing is performed using the transmitted encoded audio information and the reproduced encoded audio information. . The detailed configuration of the variable rate speech decoder 4 will be described later.

Next, a second embodiment of the present invention will be described. FIG. 2 is a block diagram showing the configuration of a variable rate speech coding system to which the channel coding function of the present invention according to claim 6 is added and its apparatus. FIG. 1A shows a variable rate audio encoding device used on the transmitting side, and FIG. 1B shows an audio decoding device on the receiving side. In the figure, 1 to 4 are the same as the same reference numerals in FIG. Reference numeral 5 is a channel encoder, and 6 is a channel decoder.

The variable-rate speech encoder 1 encodes the input speech signal a3 sampled at 8 kHz and quantized with an accuracy of 12 bits or more, and the resulting encoded speech information has a large temporal variation. The coded voice information c3 of the class selected based on the transmission rate control signal b3 is output.

Here, the transmission rate control signal is output from the channel state estimating means (one component of the adaptive modulation system) for estimating the channel state on the receiving side according to the reception level and the like and determining the optimum transmission capacity. The control signal is a signal for designating a transmission rate to the variable rate speech encoding device and the decoding device of the present invention every predetermined unit time (corresponding to the frame length: for example, 5 msec). The detailed configuration of the variable rate speech encoder 1 will be described later.

The channel encoder 5 is for encoding encoded voice information c3.
Is subjected to channel coding processing at a rate based on the transmission rate control signal b3, and error-protected coded speech information d3 is output. As a communication channel coding method, a redundancy code check method (CRC: Cyclic Redundancy Check) generally used for error detection and correction of data transmission, or convolutional coding is used. The variable rate transmission frame buffer 2 is provided with error-protected encoded voice information d3.
Is temporarily stored and error-protected coded voice information e3 is output to the adaptive modulation means side at a rate based on the transmission rate control signal b3.

The variable rate reception frame buffer 3 of the variable rate audio decoding device used on the receiving side in FIG.
The error-protected encoded voice information f3 from the variable-rate voice encoder on the transmission side is input from the adaptive demodulation means side, temporarily accumulated, and error-protected based on the rate specified by the transmission rate control signal g3. The encoded encoded audio information h3 is output. Here, the transmission rate control signal g3 is the same as the transmission rate control signal b3 referred to by the variable rate audio encoding device on the transmission side.

The channel decoder 6 receives the encoded voice information h3 which is error-protected, and is based on the rate designated by the transmission rate control signal g3, which is a generally used method such as CRC or Viterbi decoding. Error detection,
The corrected and channel-decoded encoded voice information i3 is output.

The variable rate speech decoder 4 decodes the encoded speech information i3 and outputs the reproduced speech signal j3. At this time, the coded voice information of the class with small temporal fluctuation, which was not transmitted in the frame indicating that the transmission rate control signal g3 is a frame of low rate transmission, was transmitted before and after that. After the frame information is interpolated or the frame information transmitted immediately before is used as it is for reproduction, a voice decoding process is performed using the transmitted coded voice information and the reproduced coded voice information. The detailed configuration of the variable rate speech decoder 4 will be described later.

FIG. 3 shows a first embodiment and a second embodiment of the present invention.
2 is a detailed block diagram of the variable rate coder 1 of the embodiment of FIG.
It is a specific configuration example of the variable rate speech encoder 1 of (A). This embodiment is based on a well-known code excitation linear prediction (CELP) speech coding system. In the figure, 11 is a linear prediction analyzer, 12 is a frame power calculator, 13 is an adaptive codebook, 14 is a noise codebook, 15 and 16 are gain adjusters, 17 is an adder, 18 is a synthesis filter, and 19 is addition. 20 is a perceptual weighted waveform distortion minimization controller, and 21 is a parameter divider / thinner.

In the figure, the linear prediction analyzer 11 receives the input speech signal a1 and outputs L for each frame length (5 msec: corresponding to a variable time interval of the transmission rate in the adaptive modulation system).
Spectral envelope information b1 such as SP (line spectrum pair: dimension is 10th order) is extracted and output to the synthesis filter 18 and the parameter divider / thinner 21. The analysis window length when obtaining the LSP is set to, for example, 5 frame lengths centering on the current frame. The frame power calculator 12 receives the input audio signal a1, calculates power information c1 for each frame, and outputs the power information c1 to the parameter divider / thinner 21 and the auditory weighted waveform distortion controller 20.

The adaptive codebook 13 is a codebook for expressing a voiced sound source waveform, and has, for example, 128 types of waveform patterns. The noise codebook 14 is a codebook for expressing unvoiced sound source waveforms, and has, for example, 512 types of waveform patterns. The adaptive codebook 13 and the noise codebook 14 output codewords d1 and e1 in accordance with the control signal n1 from the auditory weighting waveform distortion controller 20, respectively.

The gain adjuster 15 and the gain adjuster 16 are
According to the control signal n1 from the perceptual weighted waveform distortion controller 20, the gain adjustment of the code words d1 and e1 is performed, and the gain adjusted signals f1 and g1 are output. The adder 17 adds the two gain-adjusted codewords f1 and g1 to produce and output a reproduced excitation signal h1, and at the same time, the adaptive codebook 13
The reproduced sound source signal h1 is input to and the contents of the adaptive codebook 13 are updated.

The synthesis filter 18 uses the spectrum envelope information b1 to add the spectrum envelope component b1 to the reproduced sound source signal h1.
Is added and the reproduced voice signal i1 is synthesized and output. The adder 19 subtracts the input audio signal a1 from the reproduced audio signal i1 to create a prediction error signal j1 and inputs it to the auditory weighted waveform distortion controller 20.

The perceptual weighted waveform distortion controller 20 adapts the adaptive codebook 13 and the noise codebook 1 so that the perceptual weighted mean squared error of the prediction error signal j1 is minimized for each frame.
The index of 4 is controlled and selected by the control signal n1, and the indexes k1 and 11 are output to the parameter divider / thinner 21, respectively. Further, the auditory weighted waveform distortion controller 20 uses the output c1 from the frame power calculator 12 so that the auditory weighted mean squared error of the prediction error signal j1 is minimized. The gain of the device 16 is controlled and adjusted by the control signal n1,
The gain index m1 is converted into a parameter divider / decimator 21.
Output to

The parameter divider / decimator 21 classifies the coded speech information into classes according to the magnitude of the temporal fluctuation (details of the class classification method will be described later), and the transmission rate control signal o1 (b4 in FIG. 1 and FIG. 2). Then, the encoded voice information p1 of the class selected based on b3) is output to the variable rate transmission frame buffer 2 or the channel encoder 5 at a variable rate. Here, Table 1 shows an example of classifying the audio coded information.

[0030]

[Table 1]

Generally, when the characteristic parameters of the speech that is the encoded speech information are arranged in the order of large temporal variation, first, the noise codebook index l1 and the gain index m1, which are the excitation information, the frame power information cl, and then the , Adaptive codebook index kl, which is pitch information, and finally spectral envelope information. Therefore, in this example, as shown in Table 1, the parameters are classified into classes A, B, and C in descending order of temporal variation. As another classification method, in Table 1, spectrum envelope information b1 (first to third order) and spectrum envelope information b1 (4
10th order) may be replaced.

Table 2 shows the setting of the variable bit rate (frame length is 5 msec) of audio information based on the transmission rate control signal.
Shown in Here, it is assumed that the transmission rate designated by the adaptive modulation means changes in three stages of high rate, medium rate, and low rate. Table 2 also shows the bit rate when redundant bits are added by the channel encoder 5 according to the second embodiment of the present invention.

[0033]

[Table 2]

FIG. 4 shows a first embodiment and a second embodiment of the present invention.
2 is a detailed block diagram of the variable rate audio decoder 4 on the receiving side of the embodiment of FIG.
3 is a specific configuration example of the variable rate speech decoder 4 of FIG. 2 (B). In the figure, 31 is a parameter separator / interpolator, 3
2 is an adaptive codebook, 33 is a noise codebook, 34 is a gain controller, 35 and 36 are gain adjusters, 37 is an adder, 38 is a synthesis filter, and 39 is a post filter.

In the figure, the parameter separation / interpolator 31 is
The encoded voice information a2, which is the output from the variable rate reception frame buffer 3 or the channel decoder 6, is input, and after separating each parameter, the transmission rate control signal b2 (see FIG. 1).
F4, g3) in FIG. 2, and when it indicates that the frame is a low-rate transmission, it indicates that a class with a small time fluctuation (class C, or classes B and C) that has not been transmitted. The encoded audio information is reproduced by interpolating the information of the transmitted frames before and after it or by using the information of the frame transmitted immediately before as it is. The details will be described later.

The adaptive codebook 32 and the noise codebook 33 have the same contents as the codebooks 13 and 14 of the encoder on the transmission side, and the transmitted indexes c2 and c2, respectively.
The code words k2 and j2 indicated by f2 are output. Gain controller 3
4 uses the gain indexes d2 and frame power information e2 of the gain adjusters 15 and 16 (FIG. 3) on the transmission side to generate gains h2 and i2. The gain adjusters 35 and 36 adjust the gain of the codewords k2 and j2 using the gains h2 and i2. The adder 37 adds the gain-adjusted codewords 12 and m2 to generate and output a reproduced excitation signal n2, and inputs the reproduced excitation signal n2 to the adaptive codebook 32, and the codebook 32.
Update the contents of.

The synthesis filter 38 adds a spectrum envelope component to the reproduced sound source signal n2 based on the spectrum envelope information g2 and synthesizes the reproduced sound signal o2. Post filter 3
Reference numeral 9 denotes a process of improving the perceptual quality of the reproduced voice signal o.
2 and outputs the reproduced audio signal p2. p2 corresponds to h4 in FIG. 1 and j3 in FIG.

Next, the operation of the variable rate speech coding system of the present invention will be described with reference to FIG. The figure (a) shows the frame number Fn, and the frame length is 5 msec. FIG. 11B shows the result of classifying the coded voice information extracted for each frame according to Table 1, and the coded voice information of classes A, B, and C is obtained for each frame. FIG. 7C shows a transmission rate control signal, and here, it is assumed that the transmission rate shown by the three-stage thick line of high rate, medium rate, and low rate is taken. FIG. 3D shows a state in which information to be transmitted corresponding to the transmission rate indicated by the transmission rate control signal is selected from classes A, B, and C according to Table 2.
Here, it is shown that the information located in the shaded area is discarded without being transmitted, and class C at medium speed and classes B and C are discarded at low speed without being transmitted. Up to this point, the variable rate speech encoder 1 of FIG.
This is the operation in the variable rate speech encoder 1 of FIG.

Next, referring to FIGS. 5 (e) and 5 (f), FIG.
The variable rate audio decoder 4 of FIG. 2 and the method of reproducing coded audio information of classes B and C that have not been transmitted, which is executed in the variable rate audio decoder 4 of FIG. (E) shows a method of interpolating by the preceding and following frames transmitted as shown by the arrow. The interpolation uses linear interpolation. For example, each parameter of class C is not transmitted in the frames F2, F3, F4, but linear interpolation is performed for each parameter by the information of the frames F1 and F5. Class B of frame F3,
Similarly, the class C of the frame F7 is also linearly interpolated by the parameters of the frames before and after that. (F) shows a method of interpolation using the information of the immediately preceding frame transmitted as indicated by the arrow as it is. For example, the frame F2
Although the parameters of class C are not transmitted in F3 and F4, the information of frame F1 is used for each parameter as it is. For the class B of the frame F3 and the class C of the frame F7, information is similarly reproduced using the information of the immediately preceding frame.

As described above, the quality of the reproduced voice in the frame in which the propagation path condition becomes poor and the transmission rate becomes low due to the adaptive modulation is improved, and it is possible to provide the speech quality which is hardly influenced by the propagation path condition. .

Next, different improvements in actual use of the above method will be described. For example, an operation becomes a problem when a bad propagation state continues for a long time and when the propagation state is bad from the beginning. In this case, since only medium speed or low speed is selected as the transmission rate, in the above method, the information of class C or class B is not transmitted, and it becomes impossible to generate the reproduced sound. The problem is that when the bad state of the propagation path continues for a long time, or when the bad state is detected from the beginning, the period until the next high-rate frame comes is N.
The problem can be solved by using the voice coding means and the decoding means capable of transmitting the coded voice information of all classes even at a rate other than the highest rate among the transmission rates of the stages.

As a method of implementation, the parameter divider / decimator (21 in FIG. 3) in the variable rate speech encoder (1 in FIG. 1) monitors the transmission rate control signal (o1 in FIG. 3), The case where the bad propagation state continues for a long time and the case where the propagation state is bad from the beginning are detected. At this time, the parameter divider / decimator 21 quantizes all the parameters at the medium rate or the low rate so that the encoded speech information of all classes can be transmitted even at medium speed or low speed, and the encoded speech information of all classes is quantized. To transmit. As a method for reducing the speed, the following 2
One can be considered. (1) The number of bit allocations for each parameter in each frame is reduced. (2) The frame length is increased without reducing the bit allocation number for each parameter. Among these, the embodiment of the bit allocation of (1) that can realize the device configuration relatively easily is shown in Table 3.
Shown in

[0043]

[Table 3]

At low rates in Table 3, for parameters marked with *, normal quantization is performed only in the first frame, and the difference from the previous frame is quantized from the next frame. This speech coding technique is then repeated until a high rate frame is detected. By doing so, although the call quality is somewhat deteriorated, it is possible to generate the reproduced voice even when the bad propagation state continues for a long time or when the propagation state is bad from the beginning.

The device of the present invention can be easily realized by one DSP (digital signal processor) chip and a simple peripheral circuit.

[0046]

As described in detail above, by carrying out the present invention, in a wireless communication system using an adaptive modulation system, reproduced voice is less affected by the state of the propagation path and the listener feels uncomfortable. Is reduced, and excellent call quality can be maintained, which has a great practical effect.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a first embodiment of a variable rate speech coding system according to the present invention.

[Fig. 2] Fig. 2 is a configuration diagram showing a second embodiment of the variable rate speech coding system of the present invention.

FIG. 3 is a diagram showing a configuration example of the variable rate speech encoder of FIGS. 1 and 2 of the present invention.

FIG. 4 is a diagram showing a configuration example of the variable rate speech codec shown in FIGS. 1 and 2 of the present invention.

[Fig. 5] Fig. 5 is an operation explanatory diagram of the variable rate audio encoding system of the present invention.

[Explanation of symbols]

 1 variable rate speech encoder 2 variable rate transmission frame buffer 3 variable rate reception frame buffer 4 variable rate speech decoder 5 channel encoder 6 channel decoder 11 linear prediction analyzer 12 frame power calculator 13, 32 adaptive Codebook 14,33 Noise codebook 15,16,35,36 Gain adjuster 17,19,37 Adder 18,38 Synthesis filter 20 Auditory weighted waveform distortion minimization controller 21 Parameter division / decimation device 31 Parameter separation / Interpolator 34 Gain controller 39 Post filter

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location H04L 27/00 H04L 27/00 Z

Claims (7)

[Claims] 1. A received signal is monitored to detect the state of a channel for each frame, a high-speed transmission rate is designated when the state of the channel is good, and the transmission rate is set to N as the state of the channel deteriorates. Communication of an adaptive modulation system provided with a propagation path state estimating means for outputting a transmission rate control signal for reducing the transmission speed in stages, and an adaptive modulation means and an adaptive demodulation means for changing the transmission rate to N stages by the transmission rate control signal. A variable-rate audio encoding system to be applied to a system, in which a transmitting side performs audio encoding processing on an input audio signal for each frame to extract encoded audio information, and then associates it with the transmission rate control signal. The coded voice information is classified into N stages from information with large temporal fluctuation to information with small fluctuation, and for the frame in which the transmission rate control signal shows the maximum rate. The encoded voice information of all classes is output to the adaptive modulation means side at a rate instructed by the transmission rate control signal, and as the rate indicated by the transmission rate control signal becomes lower, the temporal variation of Variable-rate audio that discards encoded audio information from a smaller class and outputs only the remaining encoded audio information of a class with large temporal fluctuation to the adaptive modulation means side at a rate indicated by the transmission rate control signal. The receiving side receives the coded voice information from the adaptive demodulation means, reproduces the coded voice information of all classes in the frame of high rate transmission based on the transmission rate control signal, and receives the low rate. The coded audio information of the class with small temporal fluctuation that was not transmitted in the transmission frame is converted into the transmitted information of the frames before and after the frame. After being interpolated or reproduced using the information of the frame transmitted immediately before, as it is, reproduced audio signal by performing audio decoding processing of the transmitted encoded audio information and the reproduced encoded audio information. A variable-rate speech encoding system comprising variable-rate speech decoding means for generating 2. The variable-rate speech coding means, when the transmission rate control signal of the N-step transmission rate output from the propagation path state estimating means is in a bad propagation path state for a long time or is bad from the beginning. When the transmission rate is other than the highest rate, which indicates the state, until the next high-rate frame is transmitted, all classes of encoded audio information are transmitted as in the case of the highest transmission rate. The variable rate speech coding system according to claim 1, wherein 3. The variable-rate speech coding means according to claim 1, wherein the variable-rate speech coding means and the variable-rate speech decoding means are constituted by a code-excited linear predictive coding method. method. 4. The N stages according to claim 1, wherein the transmission rate is set to three stages of high speed, medium speed, and low speed, and the coded voice information is classified into classes according to the transmission rate. In order of increasing dynamic variation, class 1 is classified into three stages: high-order component of spectrum envelope information, class 2 is low-order component of spectrum envelope information and pitch information, and class 3 is frame power information and sound source signal information. The variable rate speech coding system according to claim 1, which is characterized in that. 5. The N stages according to claim 1, wherein the transmission rate is set to three stages of high speed, medium speed, and low speed, and the coded voice information is classified into time by classifying the coded voice information corresponding to the transmission rate. In order of increasing dynamic variation, class 1 is classified into three stages: low-order component of spectrum envelope information, class 2 is high-order component of spectrum envelope information and pitch information, class 3 is frame power information and sound source signal information. The variable rate speech coding system according to claim 1, which is characterized in that. 6. The variable rate audio coding system according to claim 1, wherein an error correction / detection function is added to the coded audio information. 7. The transmission rate of the channel is changed by monitoring the received signal, detecting the state of the channel for each frame, and changing the modulation multi-value number in N stages for each frame in accordance with the state of the channel. A propagation path state estimating means for outputting a transmission rate control signal for designating a high transmission rate when the state is good and for reducing the transmission rate to N stages as the state of the propagation path deteriorates, and the transmission rate control signal. A variable-rate voice coding system for application to a communication system of an adaptive modulation system provided with adaptive modulation means and adaptive demodulation means for changing the transmission rate to N stages by The encoded audio information is extracted for each frame by extracting the encoded audio information, and then the encoded audio information is distributed from information having a large temporal variation to information having a small variation corresponding to the transmission rate control signal. Classify into N stages and output the coded voice information of all classes to the adaptive modulation means side at the rate indicated by the transmission rate control signal for the frame in which the transmission rate control signal shows the highest rate. However, as the rate indicated by the transmission rate control signal becomes lower, the coded speech information is discarded from the class having the smaller temporal variation according to the degree, and the remaining coded speech of the class having the large temporal variation. A variable rate audio encoding means for outputting only information to the adaptive modulation means side at a rate indicated by the transmission rate control signal is provided, and the receiving side receives the encoded audio information from the adaptive demodulation means, and outputs the transmission rate. Based on the control signal, the coded audio information of all classes is reproduced in the frame of high rate transmission, and the time that is not transmitted in the frame of low rate transmission. Codes that have been transmitted after interpolating the coded voice information of a class with small fluctuations with the information transmitted in the frames before and after the relevant frame, or after reproducing using the information in the frame transmitted immediately before the frame as it is A variable-rate audio encoding method comprising variable-rate audio decoding means for producing a reproduced audio signal by audio-decoding the encoded audio information and the reproduced encoded audio information.