JPH08223125A - Sound decoding device - Google Patents

Abstract

PURPOSE: To reduce a sense or incompatibility that the tone color of the pseudo background noise in a silent section is unnatural when quick disturbance in one frame occurs just before the stop of transmission due to the change from a sound section to the silent section in the case of reception and decoding of a signal subjected to adaptive difference PCM encoding. CONSTITUTION: A demodulated signal (a) from a reception demodulator 2 is demultiplexed to encoded data (c) and a sound detection flag (b) indicating the presence/absence of sounds by a demultiplexer 3. A prediction coefficient (g) used to decode the encoded data (c) in an ADPCM decoder 5 is extracted by a prediction coefficient holder 6 and is averaged in each frame and is given to an auxiliary prediction coefficient holder 7, and this holder 7 always successively calculates the average value of a prediction coefficient average value g' in plural frames and holds it. When changing to the silent section, the held average value in plural just preceding frames is given to the ADPCM decoder 5 to decode the pseudo background noise.

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, and more particularly to a receiving side speech decoding apparatus in speech coding communication in which an adaptive differential PCM system is adapted to a coding system.

[0002]

2. Description of the Related Art Adaptive differential pulse code modulation (ADPCM:
The Adaptive Differential Pulse Code Modulation) method is a high-efficiency speech coding method in which redundancy is compressed. It uses adaptive prediction that predicts the current speech waveform from past speech waveforms by using the correlation of speech signals, and adaptive quantization that changes the quantization step size according to the predicted speech waveform size. .

When voice communication is performed, it is said that the time rate during which either one is uttering is about 35%. In recent years, the range of personal communication, which is a communication mainly based on individuals, has been expanding. There, mainly voice communication using a portable terminal is used. As a matter required for such a mobile terminal, the first is cordless. Secondly, it is required to reduce the circuit power consumption because a battery is used for convenience of carrying and it is required to endure long-term use. As a method of reducing the circuit power consumption more than ever before, there is a method of paying attention to the speech utterance time rate, operating the transmission circuit only when the utterance is made, and putting the circuit in a rest state during other transmission times. In order to realize such a technique, a voice detection function may be provided on the transmission side and a discontinuous transmission device corresponding to the presence or absence of voice may be added. In that case, the problem is on the receiving side. That is, the reproduced sound is intermittent on the receiving side, which makes the sound very unpleasant. The cause is that the voice and the background noise are superposed when transmitting the voice, but when the voice is not transmitted, the background noise is not transmitted, and the background noise is modulated and transmitted only when the voice signal is continuous. It is known to be for this reason.

As a method of solving such a problem,
A method is known in which a pseudo noise similar to the background noise on the transmitting side is generated on the receiving side in a section where a voice signal is not transmitted from the transmitting side. An example will be described. FIG. 3 is a block diagram of a conventional ADPCM audio decoding apparatus.
1 is an antenna. Reference numeral 2 denotes a reception demodulator, which receives and demodulates a modulated wave in which a voice detection flag of a voice detector for judging the presence / absence of voice transmitted from the transmission side and ADPCM coded data are multiplexed and modulated. A demultiplexer 3 demultiplexes the received and demodulated signal into ADPCM coded data and a voice detection flag. Reference numeral 4 denotes a controller, which receives a voice detection flag from the demultiplexer 3 and receives an ADPCM decoder,
It is a controller that outputs a control signal for performing the control described later to each of the prediction coefficient holders. 5 is ADPCM
A decoder, which receives a control signal e from the controller 4, resets, generates random ADPCM data for pseudo background noise, transmits / receives data to / from the prediction coefficient holder, and performs ADPCM encoding from the demultiplexer 3. Decrypt the data. 9
Calculates the average value for each frame by extracting the prediction coefficient which is an internal variable of the ADPCM decoder 5, stores the average value by the control signal from the controller 4, and stores the value in ADPCM.
It is a prediction coefficient holder provided to the decoder 5. Reference numeral 8 is a speaker.

Next, the operation of the conventional circuit shown in FIG. 3 will be described. The modulated wave in which the voice detection flag and the ADPCM encoded data are multiplexed and modulated is received by the antenna 1, received and demodulated by the reception demodulator 2, and sent to the demultiplexer 3 as a demodulation signal a.
The voice detection flag mentioned here is a flag indicating the result of detection by the voice detector of a portion (voiced section) of the input voice of the transmission side encoder and a portion (voiceless section) of the input voice.
The demultiplexer 3 separates into a voice detection flag b and ADPCM encoded data c. In this case, the ADPCM encoded data c has 5 msec as one frame. The voice detection flag b is sent to the controller 4 every 5 msec. When the controller 4 receives the voice detection flag b and shifts from the voiced section to the silent section, it outputs a control signal d indicating this. ADP
Since the modulation signal is interrupted in the silent section, the CM decoder 5 internally generates random data within a range that ADPCM encoded data can take, and decodes and outputs the data as pseudo background noise. The prediction coefficient holder 9 uses the ADPC in order to add the spectrum information of the actual noise to the pseudo background noise generated by the ADPCM decoder 5 when transmission is interrupted.
The second- and sixth-order prediction coefficients g extracted from the M decoder 5 are calculated by calculating an average value for each frame, updated and held, and the control signal d
Therefore, the average value f of the prediction coefficients of the last frame in the voiced section
Is input to the ADPCM decoder 5 to give pseudo background noise. The speech detector on the encoder side has a number of about several tens in order to prevent the ending of conversation when changing from a voiced section to a silent section.
Since the function generally called hangover for determining a voiced section is provided for 0 msec, the prediction coefficient value of the last frame of the voiced section is the value of noise superimposed on the voice. Therefore, even when decoding is performed using random data, noise (pseudo noise) having a tone color similar to the noise superimposed on the actual voice on the encoder side is generated.

[0006]

Here, FIG. 4 is a time chart of signals of respective parts, and FIG. 4 (x) shows a quadratic prediction coefficient a as an example of prediction coefficients for voice + background noise.
It shows a time change of 1 (t). Voice detection flag (b) to add spectrum information to pseudo background noise
The average value of the last section y (5 msec, 1 frame) in which the voice is present is used to give spectrum information to the subsequent pseudo background noise.
In addition, when a rapid change of the disturbance is added to the above, if the average of the coefficient for this section is used, a value apart from the prediction coefficient value of the noise superimposed on the voice is used as shown in (z). The noise becomes a tone color different from the actual background noise, and the listener feels uncomfortable. That is, in the conventional circuit configuration, there is a drawback in that the section A (when there is no disturbance) and the section C (when there is disturbance) in FIG.

The object of the present invention is to reduce the adverse effect on the pseudo background noise generation in the silent section on the receiving side when there is a rapid disturbance immediately before the silent section, which is a problem of the prior art, and to perform ADPCM speech coding. An object of the present invention is to provide a speech decoding device that effectively inserts pseudo background noise into a silent section in the system.

[0008]

A speech decoding apparatus of the present invention comprises a reception demodulator for receiving and demodulating coded data and a speech detection flag which are transmitted only when there is speech in adaptive differential PCM coding communication. A demultiplexer that separates the received demodulated signal into encoded data and a voice detection flag; a control unit that receives a voice detection flag from the demultiplexer and outputs a control signal corresponding to the presence or absence of voice; In the speech decoding device, the encoded data from the decoder is decoded frame by frame to output a reproduced speech signal, and in the silent section, an internally generated random data is decoded and output as pseudo background noise. A prediction coefficient holder that extracts the prediction coefficient used by the ADPCM decoder for decoding, calculates an average value for each frame, stores and updates it, and outputs
The average value of the prediction coefficient output from the prediction coefficient holder for each frame is stored for several frames, and the frame average value for several frames is calculated and updated and stored every time one frame is stored, and the control is performed. An auxiliary prediction coefficient holder for giving the frame average value updated and stored to the ADPCM decoder when the signal changes from the voiced section to the silent section.

[0009]

1 is a diagram showing an example of the configuration of a speech decoding apparatus according to the present invention. In FIG. 1, 1 is an antenna. Reference numeral 2 denotes a reception demodulator, which is a voice detection flag and ADP of a voice detector that determines the presence / absence of voice transmitted from the encoder on the transmission side.
A modulated wave that is multiplexed with the CM encoded data and modulated is received and demodulated. A demultiplexer 3 demultiplexes the received and demodulated signal a into ADPCM coded data c and a voice detection flag b. Reference numeral 4 denotes a controller, which receives the voice detection flag b from the demultiplexer 3 and outputs to the ADPCM decoder 5 and the auxiliary prediction coefficient holder 7 control signals for performing the control described later. An ADPCM decoder 5 receives a control signal e from the controller 4, generates random ADPCM data for pseudo background noise, transmits / receives data to / from a prediction coefficient holder, and an auxiliary prediction coefficient holder, and a demultiplexer. A from 3
Decode the DPCM encoded data c. 6 is ADPCM
A prediction coefficient g which is an internal variable of the decoder 5 is extracted, an average value is calculated for each frame, the average value g ′ is stored and updated, and the value g ′ is given to the auxiliary prediction coefficient holder 7. It is a coefficient holder. 7 stores and updates the value g'calculated by the prediction coefficient holder 6 in the memory that stores several frames from the newer one, by receiving the control signal d of the controller 4, It is an auxiliary prediction coefficient holder that inputs the frame average value i of the prediction coefficient to the ADPCM decoder 5 in order to calculate the frame average value i and give pseudo background noise. Reference numeral 8 is a speaker.

The operations other than the operations of the prediction coefficient holder 6 and the auxiliary prediction coefficient holder 7 are the same as those in the prior art, and only the differences from the prior art will be described. The prediction coefficient holder 6 uses ADP
The prediction coefficient g calculated in the CM decoder 5 is extracted, the average value is calculated for each frame, the data is held and updated, and the calculation result g ′ is output to the auxiliary prediction coefficient holder 7. The auxiliary prediction coefficient holder 7 stores and updates the average value g ′ for each frame calculated by the prediction coefficient holder 6 in a memory capable of storing several frames from a new one, and updates and updates one frame of data. Each time, the frame average value of the prediction coefficient is calculated from the entire data, and when the control signal d indicating that the silent section by the voice detection flag is set is received from the controller 4,
The frame average value i of the prediction coefficient calculated at this point is input to the ADPCM decoder 5 in order to give the spectrum information of the actual background noise to the pseudo background noise generated in the silent section.

FIG. 2 is a detailed structural example of the auxiliary prediction coefficient holder 7 of FIG. In FIG. 2, 11 is a first prediction coefficient memory for storing and updating the prediction coefficient values for several frames from the newest one. 12 is the first prediction coefficient memory 1
The frame average value is stored in the prediction coefficient memory 11 for one frame from the prediction coefficient values for several frames stored in 1.
It is an average value calculator that calculates each time it is updated. Reference numeral 13 stores and updates the frame average value calculated by the average value calculator 12, and when the control signal d changes from the voiced section to the silent section, outputs several frame average values i of the prediction coefficient to the ADPCM decoder 5. 2 is a second prediction coefficient memory.

Next, the operation of the auxiliary prediction coefficient holder 7 will be described. The first prediction coefficient memory 11 is a memory that holds and updates the average value g ′ in frame units held in the prediction coefficient holder 6 for several frames. Eg 5
It is a memory that can store frames, and always stores 5 frames from the latest value. Then, each time one frame is newly stored, the average value calculator 12 calculates a frame average value from the data for five frames at that time. Then, this frame average value is sent to the second prediction coefficient memory 13. The second prediction coefficient memory 13 stops the memory update when the silent period is reached by the control signal d, and outputs the frame average value i immediately before the silent period. The ADPCM decoder 5 uses this value i to provide the spectrum information to the pseudo background noise.

[0013]

As described in detail above, a prediction coefficient which is averaged and held over several frames is given to the pseudo background noise, so that a signal or the like which changes abruptly in a short time within one frame is input. Also in the case, the pseudo background noise having substantially the same spectrum information as the background noise superimposed on the voice can be generated without being directly affected by the signal.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration example of a speech decoder according to the present invention.

FIG. 2 is a configuration diagram of an auxiliary prediction coefficient holder of FIG. 1 according to the present invention.

FIG. 3 is a diagram showing a configuration example of a conventional speech decoder.

FIG. 4 is a diagram showing an example of temporal change of each signal.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Antenna 2 Reception demodulator 3 Demultiplexer 4 ADPCM decoder 5 Controller 6 Prediction coefficient holder 7 Auxiliary prediction coefficient holder 8 Speaker 9 Prediction coefficient holder 11 First prediction coefficient memory 12 Average value calculator 13 Second Prediction coefficient memory a Demodulated signal b Voice detection flag c ADPCM coded signal d Control signal e Control signal f Prediction coefficient value given to pseudo background noise (average value in frame units) g Prediction coefficient h Reproduced speech signal i Pseudo background noise Prediction coefficient value to give (average value of several frames)

Claims (1)

[Claims] 1. A reception demodulator for receiving and demodulating coded data and a voice detection flag which are transmitted only when there is voice in adaptive differential PCM coded communication, and a coded data and voice detection flag for the received demodulated signal. Demultiplexer for demultiplexing, a control unit for receiving a voice detection flag from the demultiplexer and outputting a control signal corresponding to the presence or absence of voice, and decoding the encoded data from the demultiplexer in frame units. And outputs the reproduced voice signal, and in the silent section, decodes internally generated random data and outputs it as pseudo background noise.
In a speech decoding device provided with a DPCM decoder, a prediction coefficient holder for extracting the prediction coefficient used by the ADPCM decoder for decoding, calculating an average value for each frame, storing and updating and outputting the prediction coefficient, and the prediction coefficient The average value of the prediction coefficient output from the holder in units of frames is stored for several frames, and the frame average value for several frames is calculated and updated and stored every time one frame is stored. A speech decoding apparatus comprising: an auxiliary prediction coefficient holder that gives the ADPCM decoder the frame average value updated and stored when a sound section is changed to a silent section.