JP3315708B2 - Voice codec with comparison attenuator - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a countermeasure against transmission errors of a voice codec in a voice coding system.

[0002]

2. Description of the Related Art In recent years, in digital communication, an application for replacing a part of a wired network represented by a current analog cordless telephone with wireless transmission has been studied. In this case, in consideration of compatibility with a wired network, it is appropriate to use a conventional speech codec, for example, an adaptive difference PCM (ADPCM) scheme. However, the occurrence of transmission errors is unavoidable in wireless transmission, and transmission errors are more likely to occur in the digital system than in the analog system, so that some measures are required. This can be reduced to some extent by a technique such as retransmission if the real-time property is ignored, but the retransmission technique cannot be applied to audio signal processing such as conversation since real-time property (delay time within 10 to 20 ms) is essential. Although it is conceivable to use an error correction code, it is not appropriate because an increase in the amount of information cannot be avoided. Therefore, if the voice codec itself can take measures against errors, its utility value is great.

[0003]

As one example, there is a speech codec with a comparison attenuator separately filed by the present inventors.
However, since no countermeasures are taken against transmission errors in the control variables, the influence of the transmission errors remains in the data at the subsequent stage, and the quality of the reproduced voice is not sufficiently improved. FIG. 4 is a waveform diagram showing a difference between error rates of a waveform of a reproduced audio signal, and shows a reproduced waveform in a 32 kbps ADPCM audio codec as an example of an influence of a transmission error. FIG. 4A shows a reproduced waveform when no error occurs, and FIG. 4B shows a reproduced waveform when an error having an error rate of 10 ^-2 occurs. In FIG. 4B, an impulse-like signal having a very large amplitude, which is not seen in FIG. 4A, is observed. The main ones are indicated by arrows in the figure. This impulse-like signal is very unpleasant to hear. FIG. 3C shows a reproduced waveform in an ADPCM speech codec with a comparison attenuator separately proposed as a measure against such a problem. Although the number of impulsive signals due to transmission errors has decreased, some remain. An object of the present invention is to reduce the effect on transmission sound of transmission errors, which is a problem of the prior art,
It is an object of the present invention to provide a speech codec with a comparative attenuator, which can reduce the deterioration of the quality of the reproduced speech of the codec.

[0004]

SUMMARY OF THE INVENTION A speech codec with a comparative attenuator according to the present invention comprises an input speech signal and one step from an adaptive predictor.
Adaptive quantization based on the difference from the prediction signal before
And sends the encoded signal to the transmission line
Both quantize scales from the quantizer scale factor adaptor
The encoded signal is dequantized according to the
Encoding the reproduced residual signal to be input to the adaptive predictor
Device and the coded signal received via the transmission path including the wireless line.
Quantization scale factor adaptation by adaptive inverse quantizer
Inverse quantization according to the quantization scale factor from the
The residual signal obtained and the prediction one step before from the adaptive predictor
And a decoder that outputs a reproduced audio signal from the measured signal
In the speech codec of the adaptive differential coding scheme, before
The encoder encodes the reproduction residual signal and the one-step previous signal.
The reproduction signal obtained by adding the prediction signal and the quantization scale
The threshold calculated according to the rule factor
A first comparator for reducing the amplitude of the reproduced signal by a predetermined value;
A first attenuator for decreasing the amplitude of the reproduced signal;
When the value is below the threshold value, the reproduced signal is
And the amplitude of the reproduced signal exceeds the threshold value.
The output of the first attenuator is used as the reproduced audio signal.
And a first switching means for switching and outputting.
Impulse with an attenuator superimposed on the input audio signal
A decoder configured to attenuate the audio signal of the
Is calculated according to the playback audio and the quantization scale factor.
A second comparator for comparing the output threshold value with said threshold value;
A second attenuator for attenuating the amplitude of raw voice by a predetermined value
And the quantization scale factor is attenuated by a predetermined value.
A third attenuator for adjusting the amplitude of the reproduced sound to the threshold.
If the value is less than or equal to the value, the playback audio is output as
And outputs the quantization scale factor as it is.
And when the amplitude of the reproduced sound exceeds the threshold value,
The output of the second attenuator is switched as the reproduced audio output.
And the output of the third attenuator is adjusted by the adaptive inverse
Switch output as quantization scale factor for the densifier
A second comparison attenuator comprising second switching means
Impulse of the encoded signal caused by transmission error
Characterized by being configured to attenuate the shape of noise
Things .

[0005]

FIG. 2 is a block diagram showing a comparative attenuator according to an embodiment of the present invention. FIG. 2A is a block diagram of a comparative attenuator provided on the encoder side. The comparator 1 compares the input reproduction signal Sr (k) with the threshold value, and
When (k) does not exceed the threshold value, the input reproduction signal Sr (k) is output as it is, and when (k) exceeds the threshold value, the reproduction signal Sr '(k) attenuated through the attenuator 2 is output. The output is switched by the switch 3 so as to perform the switching. FIG.
(B) is a block diagram of a comparison attenuator for the decoder side. The threshold value and the reproduced signal Sr (k) are compared by the comparator 1 and, based on the result, the reproduced signal Sr (k) and the attenuator 2 are used as they are.
And the reproduced signal Sr '(k) passed through the switch 3 and output. At this time, the switching unit 21 outputs the control variables that have passed through the attenuator 20 at the same time.
Is controlled. As described above, the amplitude of the finally output reproduced signal output is limited to within a predetermined level, and the decoder also modifies the control variables. By incorporating this comparison attenuator into the speech encoder and the codec, a speech codec with a comparison attenuator is constructed. An encoder attenuates an impulse sound signal to a certain level to make it easier to detect impulse-like noise caused by a transmission error in a decoder. In this case, the impulse sound signal is suppressed by the encoder. However, such a case is rare in general speech sound, and the difference caused by adjusting the attenuation of the comparative attenuator is small. There is almost no deterioration in the quality of the reproduced sound due to the provision of the comparative attenuator of the present invention.

Next, as an embodiment of the present invention, ADPCM
A case in which the above-described comparison attenuator is incorporated in a speech codec based on the system will be described. FIG. 1 is a block diagram of an ADPCM speech codec with a comparative attenuator according to the present invention. The processing flow is shown below. First, the encoder calculates the difference (residual signal a) between the input audio signal and the reproduced signal one step before, that is, the prediction signal f. The input residual signal a is adaptively quantized according to the level by the adaptive quantizer 4 and transmitted as a coded output c. The quantization width used at that time, that is, the quantization scale factor b is derived by the quantization scale factor adaptation unit 5. An adaptive inverse quantizer 6 reproduces and outputs a reproduction residual d based on the value (encoded output c) quantized by the adaptive quantizer 4. The reproduction signal e is reproduced using the reproduction residual d and the prediction signal f one step before. The reproduced signal e is input to the above-mentioned comparison attenuator 8,
When the value is equal to or larger than the threshold value, the value of the reproduced sound e is attenuated and the reproduced signal e 'is output. The reproduced voice e 'is also input to the adaptive predictor 7 to generate a prediction signal f used for prediction in the next step. The difference between the predicted signal f output from the adaptive predictor 7 and the input speech in the next step is quantized by the above procedure. This is the flow of processing on the encoder side.

On the decoder side, on the other hand, the coded output c is transmitted from the coder and the quantization scale factor adaptor 9 determines the quantization width, that is, the quantization, based on the code c ′ affected by the radio channel. Calculate and output the scale factor g.
The adaptive inverse quantizer 10 decodes the residual h using the quantization scale factor g ′ and the code c ′, which have been processed by the comparison attenuator 12 in the previous step, as inputs. The adaptive predictor 11 outputs a prediction signal i using the decoded residual h and the reproduced speech j ′ one step before. The reproduced signal j is synthesized by adding the prediction signal i and the reproduction residual h. The reproduced sound j is input to the above-described comparison attenuator 12, and when the reproduced sound j is equal to or larger than the threshold value, the reproduced sound j is attenuated and output assuming that a transmission error is included. This is the reproduced audio signal j '. In the next step, the quantization scale factor g used as a control variable is also attenuated and output as the quantization scale factor g '. Like the encoder, the reproduced speech j ′ is input to the adaptive predictor 11 to generate a prediction signal i used for prediction in the next step. The quantization scale factor g ′ is input to the adaptive inverse quantizer 10 and used for inverse quantization in the next step. Is performed. FIG. 3 is a block diagram of a comparative attenuator in which a numerical value of a specific example is entered in a main part of the present invention. FIG. 3A shows the encoder side, and FIG.
Is a block diagram of the comparison attenuator inserted on the decoder side
It is. In the case of an adaptive encoding method such as the ADPCM method, it is better to determine the threshold value from a variable having a large relation with the reproduced sound, such as a quantization scale factor, instead of making the threshold unchanged.

[0008] As a specific example, the results of implementation when the threshold value is determined based on the quantization scale factor will be described.
FIG. 5 is a waveform diagram showing the effect of the present invention, and is 32 kbps.
9 shows the results of implementing an ADPCM speech codec with a comparative attenuator. FIG. 5 (A) shows the result when a comparative attenuator which does not process the control variable is used, and FIG. 5 (B) shows the result when the comparative attenuator of the present invention is inserted. The threshold value at this time is set to 100 using the quantization scale factor Y (k).
× 2 ^{Y (k),} and the coefficient of attenuation was set to 0.9 for both the reproduced voice and the quantization scale factor. As shown in the drawing, the influence of transmission errors is reduced by providing the comparative attenuator of the present invention which also processes the control variables in the reproduced voice waveform. Quantitatively, the present invention was able to achieve about 0.5 dB improvement in segmental SNR. Impulsive crackling discomfort was also reduced in hearing.
Furthermore, the processing amount of the present invention is as small as 0.14 MIPS,
It could be realized with about 3% of the processing amount (about 4.54 MIPS) of the ADPCM codec. In the above embodiment, AD
Although the speech codec of the PCM method has been described, the present invention can be applied to a speech codec of another method.

[0009]

As described above in detail, by incorporating the comparison attenuator of the present invention into the speech codec, the impulse noise generated when a transmission error occurs is reduced, and the quality of the reproduced speech is expressed by the segmental SNR. It can be improved by about 6%. This enables higher quality communication in a system having a wireless section in which transmission errors easily occur. Also,
Since the processing amount is also extremely small, the increase in power consumption is small even if a comparison attenuator is added, which is extremely effective for a wireless device requiring low power consumption.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing a main part of the present invention.

FIG. 3 is a block diagram showing a main part of the present invention.

FIG. 4 is a reproduction waveform diagram of a 32 kbps ADPCM voice codec;

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

[Explanation of symbols]

 Reference Signs List 1 comparator 2, 20 attenuator 3, 21 switcher 4 adaptive quantizer 5, 9 quantization scale factor adaptor 6, 10 adaptive inverse quantizer 7, 11 adaptive predictor 8, 12 comparative attenuator

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-2-239300 (JP, A) JP-A-60-173600 (JP, A) JP-A-63-81400 (JP, A) JP-A-1- 245606 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 19/00-19/04 H03M 7/30 H04B 14/04

Claims (1)

(57) [Claims] 1. An input speech signal and one step from an adaptive predictor.
Adaptive quantization based on the difference from the prediction signal before
And sends the encoded signal to the transmission line
Both quantize scales from the quantizer scale factor adaptor
The encoded signal is dequantized according to the
Encoding the reproduced residual signal to be input to the adaptive predictor
Device and the coded signal received via the transmission path including the wireless line.
Quantization scale factor adaptation by adaptive inverse quantizer
Inverse quantization according to the quantization scale factor from the
The residual signal obtained and the prediction one step before from the adaptive predictor
And a decoder that outputs a reproduced audio signal from the measured signal
In the speech codec of the adaptive difference encoding method, the encoder is configured to determine whether the reproduction residual signal is
A reproduction signal obtained by adding the prediction signals of
Compare with the threshold calculated according to the scale factor.
A first comparator and an amplitude of the reproduction signal by a predetermined value.
A first attenuator for attenuating, the amplitude of the reproduced signal being
If the signal is below the threshold, the reproduced signal is
And the amplitude of the reproduced signal exceeds the threshold.
The output of the first attenuator is referred to as the reproduced audio signal.
First comparing means comprising first switching means for switching and outputting.
An impulse equipped with an attenuator and superimposed on the input audio signal
And the decoder comprises a reproduction audio and the quantization scale factor.
Second comparator for comparing a threshold value calculated according to
And a second attenuating the amplitude of the reproduced sound by a predetermined value.
Attenuator and the quantization scale factor are set to predetermined values.
And a third attenuator for attenuating the amplitude of the reproduced sound.
If the threshold value is equal to or less than the threshold value, the reproduced audio
Output and quantize the scale factor
Output and the amplitude of the reproduced sound exceeds the threshold.
When the output of the second attenuator is used as the reproduced audio output
The output of the third attenuator is switched and the output of the third
Cutoff as the quantization scale factor for the inverse inverse quantizer.
A second comparison attenuator comprising a second switching means for performing a replacement output
And the input of the encoded signal caused by a transmission error is provided.
It is specially designed to attenuate pulse noise.
A voice codec with a comparison attenuator.