JPH06334613A - Voice decoder - Google Patents

Abstract

PURPOSE:To reduce unpleasant feelings of a reproduced voice due to a transmission error generated in a radio section at the time of receiving and decoding an encoded voice signal through the radio section in an adaptive difference PCM system. CONSTITUTION:A received signal (g) demodulated by a reception demodulator 8 is divided into control data (i) and encoded data (h) by a multiplex separator 8. A control part 10 detects a frame whose level is rapidly changed and the frame with many error bits, and outputs erroneous frame flags (k), (p), and (t). A predicted coefficient holder 15 always updates and stores a predicted coefficient (s), and applies a predicted coefficient mean value (r) of the previous frame to a decoder 11 when the error flag (p) is applied in order to allow the decoder 11 to operate a decoding processing. A power corrector 13 updates and stores power information (m) in each frame from a power calculator 12, corrects the power of the erroneous frame so as to be equal to the power of the previous frame when the error flag (k) is applied, and outputs the reproduced voice.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech decoding apparatus used for speech coding communication, and more particularly to an adaptive differential PCM (ADPCM: adaptive dif) for speech coding system.
The present invention relates to a voice decoding device used in a communication system to which the ferential pulse code modulation) method is applied.

[0002]

2. Description of the Related Art In recent years, the range of personal communication, which is an individual-based communication, has been expanding. There, mainly voice communication using a portable terminal is used. Items required for such a mobile terminal include cordlessization and digitization of voice. With respect to the audio digitizing system, the 64 kbps PCM system is used in the wired system, and the 32 kbps ADPCM system is also currently used. In the wireless system, the frequency band is finite, and it is better that the number of bits transmitted per channel is smaller in order to effectively use it. Furthermore, since it is shared with the wired system of the terminal, the 32kbps ADPCM system used in the wired system. Tend to be used. The 32 kbps ADPCM system has good reproduced voice quality if the error rate is sufficiently low as in a wired system, but in a transmission system including a wireless section such as a cordless telephone, occurrence of a transmission error in the wireless section is unavoidable. One of the causes of transmission error is low speed fading of several Hz. Under low-speed fading, most of the frames of transmitted data are erroneous, which causes deterioration of reproduced voice quality.

[0003]

There are several tens of burst errors in which most of the frames of the transmission data are erroneous.
If it exceeds msec, it will give an unpleasant sensation that impact sound is mixed in the reproduced sound. Although some countermeasures against such burst errors are known, no countermeasure against a system including a wireless line such as a cordless telephone is yet known. FIG. 5 is a waveform diagram showing a waveform of a reproduced voice signal in the 32 kbps ADPCM system for 5 seconds as an example of the influence of the transmission error due to the low speed fading. In this example, the ADPCM encoded data is 5
One frame is msec. FIG. 5 (A) shows a voice signal waveform when there is no error. (B) represents the number of error bits generated in one frame, and an error of about 70 bits at maximum is observed. (C) is a flag set in the error frame, and (D) is an example of the output waveform of the decoder when a transmission error occurs. The error conditions are Rayleigh fading, fading pitch 4 Hz, average error rate 1.3 × 10.
^-3 . It is observed in FIG. 5D that a signal (a) having a very large amplitude, which is not seen in (A), is superimposed. This large amplitude portion corresponds to the above-described impact sound and causes discomfort. An object of the present invention is to reduce the influence of a transmission error in a voice coding communication system including a wireless transmission section such as a cordless telephone, which is a problem of the above-mentioned conventional technology, on reproduced sound, and to improve the quality of reproduced sound wirelessly transmitted. The present invention provides a speech decoding device that reduces the deterioration of.

[0004]

A speech decoding apparatus according to the present invention includes a reception demodulator for receiving a radio wave in which a wireless carrier wave is modulated by an adaptive differential PCM speech coded signal and outputting a demodulation signal, and the demodulation signal A demultiplexer that separates and outputs encoded data and control data, and an ADPC that performs adaptive differential PCM decoding of the encoded data from the demultiplexer and outputs a decoded signal
A voice decoding device including an M decoder
The prediction coefficient used in the decoding process in the CM decoder is the ADP
A prediction coefficient given to the ADPCM decoder by taking out from the CM decoder, updating and storing the average value for each frame, stopping the updating of the storage when an error frame flag is input from the outside, reading the prediction coefficient of the immediately preceding frame, A cage,
The decoded signal output from the ADPCM decoder is output as a reproduced audio signal, the power of each frame of the decoded signal is updated and stored, and when an error frame flag is input from the outside, the update of the storage is stopped and When the power level of the error frame is corrected to be equal to the power of the immediately preceding frame and output as a reproduced voice signal, and the reception level of the reception demodulator is calculated and monitored, and when the reception level in a frame unit suddenly changes. Gives an error frame flag to the power corrector, and when the control data from the demultiplexer is checked frame by frame to detect an error bit, A
By giving an error frame flag to the DPCM decoder and the prediction coefficient holder, the prediction coefficient of the immediately preceding frame output from the prediction coefficient holder is input to the ADPCM decoder, and the ADPCM decoder outputs the prediction coefficient. And a controller that operates so as to perform a decoding process using the prediction coefficient of the frame.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing the overall configuration of a communication system including a speech decoding apparatus according to the present invention. (A) shows a coding apparatus on the transmitting side and (B) shows a decoding apparatus. FIG. 4 is a timing chart of signals at various parts of the circuit of FIG. The configuration and operation of the present invention will be described with reference to FIGS. The English small letters (each signal, flag, etc.) appearing in the description correspond to FIGS. 1 and 4.

First, the configuration of the encoding device (A) will be described. In FIG. 1A, 1 is a microphone. 2 is an ADPC for ADPCM-encoding the audio signal a
It is an M encoder. Reference numeral 3 denotes a multiplexer which multiplexes the ADPCM-coded data b and the control data c from the control data generator 5 into a frame and performs speed conversion. 5
Is a control data generator for supplying the control data c to the multiplexer 3. Reference numeral 4 is a modulation transmitter for transmitting a transmission wave e, which is obtained by modulating a carrier wave using the multiplexed data d as a modulation signal, from an antenna 6. Next, the operation of the encoding device (A) will be described. The audio signal a taken from the microphone 1 is ADPCM encoded by the ADPCM encoder 2. For example, if the sampling frequency is 8 kHz, 4
Bit data is 5 msec for one frame. AD
The PCM-encoded data b is multiplexed with the control data c from the control data generator 5 by the multiplexer 3, and the speed is converted. The speed-converted multiplexed data d is modulated by the modulation transmitter 4 and transmitted from the antenna 6.

Next, the configuration of the decoding device of (B) showing the embodiment of the present invention will be described. In FIG. 1 (B), 7
Is an antenna. 8 is the transmission wave f transmitted from the transmission side
Is a reception demodulator that receives and demodulates by the antenna 7. Reference numeral 9 denotes a demultiplexer having a function of speed-converting and demultiplexing the multiplexed data g output from the reception demodulator 8 into control data i and ADPCM-encoded data h. Reference numeral 10 denotes a controller, which checks the error of the control data i separated and output by the demultiplexer 9 and detects the error frame flags t and p when the data frame containing the transmission error is detected, respectively, by the ADPCM decoder 11 and the prediction coefficient. The function of outputting to the holder 15 and the level of the demodulated signal of the demodulated signal for each data frame are monitored, and the level of the current data frame is extremely changed from the level of the immediately preceding data frame. Also in this case, it has the function of outputting the error information k to the power corrector 13 and the function of generating the frame pulse 1 for each frame. Reference numeral 15 is a prediction coefficient holder which is an essential part of the present invention, and is an ADPCM decoder 11
The prediction coefficient s used in the decoding process for determining the tone color of the reproduced voice is always stored by temporarily calculating the average value for each frame, and when the error frame flag p from the controller 10 is received, the storage is stopped. And the ADPCM decoder 1
At the time of data decoding processing of an erroneous frame at 1, the prediction coefficient r of the immediately preceding frame having no error is read from the memory and output to the ADPCM decoder 11. Then, it has a function of restarting the storage of the memory again when the error frame flag p is not received.

Reference numeral 11 is an ADPCM decoder,
ADPCM decoding of the M-encoded data h, and the controller 1
When a frame with an error is decoded by the error coding flag t from 0, it has a function of decoding by the prediction coefficient r from the prediction coefficient holder 15. Reference numeral 12 denotes a power calculator having a function of calculating the power of each frame of the ADPCM-decoded decoded speech j and outputting the power value (power information m) to the power corrector 13. A power corrector 13 has a function of storing the power information m from the power calculator 12, and when the controller 10 gives an error frame flag k and a frame pulse l, a frame containing a transmission error. Then, the power of the decoded speech of a predetermined number of frames following that is corrected to be the same as the power of the decoded speech of the frame before the error is input, and the speaker 1 is used as the output speech.
4 has the function of outputting.

Next, the operation of the decoding device will be described.
The received wave f that has passed through the wireless section is input to the reception demodulator 8 via the antenna 7 and demodulated. The demodulated multiplexed data g is subjected to speed conversion by a demultiplexer 9 and separated and output as control data i and ADPCM coded data h. The controller 10 calculates and monitors the reception level of the demodulator 8, sets an error frame flag k to the frame whose level has changed significantly from the level of the previous frame, and checks the error of the separated control data i to make an error. The error frame flags p and k are set to the determined frame and output. Also, a frame pulse 1 indicating a frame delimiter is output. Figure 4
The reception level x of indicates that the level of the section including the transmission error is lowered as an example. The ADPCM coded data h is ADPCM-decoded by the ADPCM decoder 11 and outputs a decoded signal j. It can be seen that the ADPCM decoded signal j in FIG. 4 has a large amplitude due to the influence of the transmission error in the section including the transmission error. The power calculator 12 calculates the power of each frame of the decoded speech j and sends the power information m to the power corrector 13. The power corrector 13 stores the power information m calculated by the power calculator 12, and according to the error frame flag k and the frame pulse 1 from the controller 10,
The power of the decoded voice of the frame having the transmission error is corrected so as to be equal to the power of the decoded voice of the frame before the error is input, and output as the output voice n. As shown in FIG. 4, the power is corrected like the power m ′ after the power correction and the output voice n, and the influence of the transmission error of the reproduced output voice is reduced.

The prediction coefficient holder 15 extracts the prediction coefficient s used by the ADPCM decoder 11 in the decoding process, calculates the average for each frame, and updates and stores it. And
When the error frame flag p is input from the controller 10, the memory update is stopped, and the average value r of the prediction coefficients of the frame before the error is input is output to the ADPCM decoder 11.
When the error frame flag p is exhausted, the storage is restarted. Upon receiving the error frame flag t from the controller 10, the ADPCM decoder 11 performs the decoding process of the erroneous frame by using the prediction coefficient average value r of the frame before the error enters. By using the prediction coefficient average value r of the frame before the error is introduced, the audible discomfort of the decoded speech is reduced. When the prediction coefficient w on the transmission side in FIG. 4 is disturbed in level as in the prediction coefficient s in FIG. 4 due to a transmission error, the above processing is performed and the prediction coefficient is disturbed as in the prediction coefficient r in FIG. It can be seen that is reduced.

FIG. 2 is a diagram showing a configuration example of the power corrector 13 which is one of the features of the decoding device of the present invention. The configuration and operation of the power corrector 13 will be described with reference to FIG. First, the configuration will be described. In FIG. 2, reference numeral 21 is the power calculator 1
It is a memory that stores the power m for each frame calculated in 2. This memory 21 has a function of stopping writing when the error frame flag k is input and restarting writing by a write stop clear signal o output from the counter 23. A buffer memory 22 matches the timing of the decoded voice j output from the ADPCM decoder 11 with the power information output from the memory 21. Reference numeral 23 denotes a counter, which receives the error frame flag k, switches the switch 24 to the output side of the memory 21 to output the contents of the memory during a frame in which an error has occurred by the frame pulse 1, and thereafter the error frame flag k disappears. Then, the switch 24 is returned to the side, and the write stop clear signal o is output to the memory 21 to restart writing.
25 is a multiplier.

Next, the operation of the power corrector 13 will be described. The decoded speech j from the ADPCM decoder 11 has the power m calculated for each frame by the power calculator 12 and stored in the memory 21, and the content of the memory 21 is updated for each frame. When the error frame flag k is input from the controller 10, the updating of the memory 21 is stopped. When the error frame flag k is input to the counter 23, the switch 2
The switch signal 4 is output, and the switch 24 connected to the normal side is switched to the side. And counter 2
3 counts the number of erroneous frames by the frame pulse 1 and then supplies the memory start signal o to the memory 21 to start updating and at the same time, returns the switch 24 to the side.
With the above operation, the multiplier 25 multiplies the output from the switch 24 and the output from the buffer memory 22, so that the decoded speech power of the erroneous frame becomes the same as the power of the frame before the error. Is corrected and output.

FIG. 3 is a diagram showing an example of the structure of the prediction coefficient holder 15, which is one of the features of the decoding apparatus of the present invention. The configuration and operation of the prediction coefficient holder 15 will be described with reference to FIG. First, the configuration of the prediction coefficient holder 15 will be described. In FIG. 3, reference numeral 26 is a prediction coefficient calculator, and the ADPCM decoder 1
It has a function of calculating the average value of the prediction coefficient s obtained by the adaptive predictor of No. 1 for each frame. Reference numeral 27 denotes a prediction coefficient memory, which stores the value q calculated by the prediction coefficient calculator 26, stops the storage of the prediction coefficient when the error frame flag p arrives, and updates the memory again when the error frame flag p is exhausted. start. While the error frame flag p is set, AD
The PCM decoder 11 has a function of outputting the average value r of the prediction coefficients of the frame before the error is entered.

Next, the operation of the prediction coefficient holder 15 will be described. The prediction coefficient calculator 26 always uses ADPC.
Given the prediction coefficient s generated in the M decoder 11,
The average value q is calculated and output for each frame. The prediction coefficient memory 27 updates and stores the average value q of the prediction coefficient for each frame calculated by the prediction coefficient calculator 26, and when the error frame information p is input from the controller 10, the update of the memory is stopped and the memory is stored. The content r of the above is output to the ADPCM decoder 11. Also, when the error frame information p does not come, the memory update is started again. When the error frame flag t is input from the controller 10, the ADPCM decoder 11 uses the output r from the prediction coefficient memory 27 to perform the decoding process of the frame with an error. By such an operation, the reproduced voice quality is improved as compared with the case where the prediction coefficient affected by the transmission error is used. FIG. 5 (E) is a waveform diagram showing the waveform of the reproduced audio signal of the present invention which has been subjected to the above processing. It is shown that the very large amplitude seen in the conventional waveform of FIG. 6D is suppressed, and the audible discomfort is reduced.

[0015]

As described above in detail, as a transmission error countermeasure, a reproduction error of a transmission error is corrected by holding a prediction coefficient for a frame having an error and performing a decoding process and correcting the power. Since the influence on the quality of the reproduced voice of the decoder can be reduced,
It is extremely effective as a voice decoding device for a system including a wireless section such as a cordless telephone using the ADPCM coding system.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a speech coding apparatus and a decoding apparatus of the present invention.

FIG. 2 is a partial detailed block diagram of the present invention.

FIG. 3 is a partial detailed block diagram of the present invention.

FIG. 4 is a signal waveform diagram of each part in FIG.

FIG. 5 is a waveform diagram illustrating the effect of the present invention.

[Explanation of symbols]

 1 Microphone 2 Encoder 3 Multiplexer 4 Modulation Transmitter 5 Control Data Generator 6,7 Antenna 8 Reception Demodulator 9 Demultiplexer 10 Controller 11 Decoder 12 Power Calculator 13 Power Corrector 14 Speaker 15 Prediction Coefficient Holder 21 Memory 22 Buffer memory 23 Counter 24 Switch 25 Multiplier 26 Prediction coefficient calculator 27 Prediction coefficient memory

Claims (1)

[Claims] 1. A reception demodulator which receives a radio wave whose radio carrier wave is modulated by an adaptive differential PCM voice coded signal and outputs a demodulated signal, and a multiplexer which separates and outputs the demodulated signal into coded data and control data. In a speech decoding device including a demultiplexer and an ADPCM decoder that adaptively differentially PCM-decodes the encoded data from the demultiplexer and outputs a decoded signal, a prediction coefficient used in a decoding process by the ADPCM decoder is calculated. Prediction given to the ADPCM decoder by taking out from the ADPCM decoder, updating and storing the average value for each frame, stopping the update of the storage when an error frame flag is input from the outside, reading the prediction coefficient of the immediately preceding frame, The coefficient holder and the decoded signal output from the ADPCM decoder are output as reproduced audio signals and the power of each frame of the decoded signal is output. And a power corrector for updating and storing, and when the error frame flag is input from the outside, stopping the update of the memory, correcting the power of the error frame to be equal to the power of the immediately preceding frame, and outputting it as a reproduced audio signal. The reception level of the reception demodulator is calculated and monitored, and when the reception level of each frame suddenly changes, an error frame flag is given to the power corrector, and the control data from the demultiplexer is checked for each frame to make an error. When a bit is detected, an error frame flag is given to the ADPCM decoder and the prediction coefficient holder, so that the prediction coefficient of the immediately preceding frame output from the prediction coefficient holder is the ADP.
A speech decoding apparatus, comprising: a controller that is input to a CM decoder and operates so that the ADPCM decoder performs a decoding process using a prediction coefficient of the immediately preceding frame.