GB2269076A - Speech coding transmission system and coder and decoder therefor - Google Patents

Abstract

A speech coding transmission system in which a coder of a differential coding system quantizes a difference between a predictive value (f) from an adaptive predictor (7) and an input speech signal by an adaptive quantizer (4) in accordance with the level of the after and sends out the resulting coded signal (c) to the transmission medium. A decoder outputs a reproduced speech signal (j) on the basis of a residual signal (h) obtained by inversely quantizing a received coded signal (c') with an inverse adaptive quantizer (10) and a predictive value (i) from an adaptive predictor (11). Both or either one of the coder and the decoder includes a comparator-attenuator (8, 12) which attenuates the amplitude of the input signal or reproduced speech signal to a value smaller than a predetermined threshold value and applies the amplitude-attenuated speech signal to the adaptive predictor (7, 11). <IMAGE>

Description

SPEECH CODING TRANSMISSION SYSTEM AND CODER AND DECODER THAU7OR The present invention relates to countermeasures for transmission errors in a speech coding transmission system.

Analog or digital data transmission has long used wire telecommunication networks, in which errors during transmission are very rare, and it has been known that the error rate is in the range of 10-4 to 10-8.

With such a technical background, speech coders and speech decoders have been developed for application to a digital data transmission system via a wire telecommunication network. For example, a speech coder or decoder of an adaptive differential PCM (ADPCM) system has been developed on the assumption of an error rate less than 10-4.

The wire telecommunication network is now partly being substituted b radio transmission for a cordless telephone system, for example. From the viewpoint of matching with the wire telecommunication network, it is appropriate, in this instance, to use a conventional speech coder or decoder such as the ADPCM system.

In radio transmission, however, occurrence of a transmission is inevitable. If the real-time transmission in the radio transmission system is ignored, the transmission errors could be reduced to some extent by repeating or like technique, but the repeating technique cannot be used in speech signal processing as of conversation voice signal because the real-time property (a delay time in the range of 10 to 20 ms) is requisite to such signal processing.

Accordingly, it is necessary to take some countermeasures for errors in the application of the speech coder or decoder to the system wherein the wire telecommunication network partly includes a radio section.

One possible method is the use of an error correcting code, but this method has a defect of an increase in the quantity of information to be transmitted. Therefore, if an appropriate countermeasure can be made by the speech coder or decoder itself against errors, its utility value will be great. However, such a technique is unknown, because the speech coder or decoder intended primarily for use in the wire telecommunication network as mentioned above has not been applied to the radio telecommunication network in the past.

An object of the present invention is to provide a speech coding transmission system and a speech coder or decoder with comparator-attenuator which lessens the influence on a reproduced speech signal by transmission errors in the radio section, thereby alleviating deterioration of the quality of the reproduced speech;;- A speech coding transmission system according to the present invention comprises a coder of a differential coding system which quantizes a difference between a predictive value from an adaptive predictor and an input speech signal by an adaptive quantizer in accordance with the level of the latter and delivers a coded signal and a decoder which outputs a reproduced speech signal on the basis of a residual signal obtained by inversely quantizing the received coded signal with an inverse adaptive quantizer and a predictive value from an adaptive predictor.The speech coding transmission system according to the present invention is characterized in that both or either one of the coder and the decoder includes the comparator-attenuator which attenuates the amplitude of the input speech signal or reproduced speech signal to a value smaller than a predetermined threshold value and provides the amplitude-attenuated speech signal to the adaptive predictor.

Embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which: Fig.1 is a block diagram illustrating an embodiment of the present invention; Fig. 2 is a block diagram for illustrating the principal part of embodiments of the present invention; Fig. 3 is a block diagram showing an embodiment of the principal part of embodiments of the present invention; Fig. 4 shows reproduced waveforms by a 32-kbps ADPCM speech coder/decoder; and Fig. 5 shows reproduced waveforms for explaining the effect of the present invention.

Fig. 2 is a block diagram of the comparator-attenuator which constitutes the principal part of embodiments of the present invention.

At first, an input reproduced signal Sr(k) and a threshold value are compared with each other by a compara tor l, and when the level of the reproduced signal Sr(k) exceeds the threshold value, a switch 3 is connected to an attenuator 2 to output therefrom an attenuated reproduced signal Sr' (k), whereas when the level of the reproduced signal Sr(k) does not exceed the threshold value, the input reproduced signal' is output intact via the switch 3. Thus, the amplitude of a reproduced signal Sr"(k) which is ultimately output is limited to a certain level range.

By incorporating the comparator-attenuator into either one or both of the speech coder and decoder, the speech coder or decoder with the comparator-attenuator is formed.

That is, also in the coder an impulse-like speech signal is attenuated in a coder to a certain level, by which an impulse-like noise caused by transmission errors can be detected with ease at a decoder. In this instance, the impulse-like speech signal is suppressed in the coder but this is very rare in ordinary speech communication.

Furthermore, adjustment of the attenuation of the comparator-attenuator does not exert any influence on the speech signal and the provision of the comparator-attenuator according to the present invention scarcely deteriorates the reproduced signal quality.

Next, a description will be given of an embodiment of the present invention in which the above-mentioned comparator-attenuator is incorporated in both of the coder and decoder of the speech coder/decoder of the ADPCM system.

Fig. 1 is a block diagram of the ADPCM speech coder and decoder with the comparator-attenuator according to an.

embodiment of the present invention. The flow of its processing is given below.

In the coder: (1) A difference between the input speech signal and a reproduced signal of an immediately preceding step, i.e. a predictive signal f is detected (that is, a residual signal a is obtained).

(2) The input residual signal a is adaptively quantized after adjustment of its level by an adaptive quantizer 4, and the quantized output is transmitted as a coded output c.

(3) The quantization values for the above quantization, that is, a quantizer-scale factor b, is produced by a quantizer-scale factor adapter 5.

(4) Based on the quantized value (i.e. the coded output) by the adaptive quantizer, a reproduced residual d is reproduced by an inverse adaptive quantizer 6 and is then output therefrom.

(5) The reproduced residual d and the predictive signal f of the immediately preceding step are used to reproduce a reproduced signal e.

(6) The reproduced signal e is applied to a comparator- attenuator 8 provided according to the present invention, and when the reproduced signal e exceeds a threshold value, the reconstructed signal is attenuated and is output as an attenuated reproduced signal e'.

(7) The reconstructed signal e' is applied to a composite adaptive predictor 7, which generates the predictive signal f which is use'for the prediction in the next step.

(8) A difference between the predictive signal f from the composite adaptive predictor 7 and the input speech signal of the next step is quantized following the above procedure.

The above is the flow of processing on the side of the coder.

'In decoder: (1) Based on'a coded output c' which is the coded output c influenced by the transmission line (a radio channel, for example) during transmission, the quantization value, i.e. a quantizer-scale factor Y(k)g is computed by a quantizer-scale factor adaptor 9 and output therefrom.

(2) The quantizer-scale factor q and the coded output c' are input into an inverse adaptive quantizer 10, wherein a residual h is decoded.

(3) The decoded residual h and a regenerated speech signal i' of the immediately preceding step are used to obtain a predictive signal i in a composite adaptive predictor 11.

(4) The predictive signal i and the decoded residual h are added together to synthesize a reproduced speech signal 2 (5) The reproduced speech signal i is input into a comparator-attenuator 12 which is a characteristic feature of the present invention, and when the'input exceeds a threshold value, the former is regarded to contain transmission errors, the reproduced speech signal i is attenuated, thereafter being output as a reproduced speech 2.

(6) As in the case of the coder, the reproduced speech signal i' is input into the composite adaptive predictor 11, which generates the predictive signal i which is used for prediction in the next step.

Fig. 3 is a block diagram of an embodiment of the comparator-attenuator wich forms he principal part of the present invention.

In the case of an adaptive coding system such as the ADPCM system, it is preferable not to hold the threshold value invariable but to determine it based on a variable closely related to the reproduced speech, such as the quantizerscale factor.

To show an example of the influence of transmission errors, Fig. 4 depicts waveforms of reproduced speech signals by 32-kbps ADPCM speech coder or decoder. Fig.

4(A) shows a waveform in case of no error, and Fig. 4(B) a waveform in a case where errors of an error rate 10-2 occurred. It is seen from Fig. 4(B) that impulsive noises of very large amplitudes (as indicated by arrows in Fig.

4(B)) unseen in Fig. 4(A) are superimposed.

Now, a description will be given of the result obtained in a case where the threshold value was determined on the basis of the quantizer-scale factor Y(k).

Fig. 5 is waveform diagram for explaining the effect of the present invention, showing the result of using the 32kbps ADPCM speech coder/decoder with the comparator-attenuator. Fig. 5(A) shows waveforms obtained in the case of the prior art without the comparator-attenuator, and Fig.

5(B) shows waveforms obtained with. an embodiment of the present invention employing the comparator-attenuator. In this case, the threshold value was set to 7x2Y(k) through utilization of the quantizer-scale factor Y(k) and the coefficient of attenuation was set to 0.9. As shown in Fig. 5, the influence of the transmission error on the reproduced speech waveform is lessened by the provision of the comparator-attenuator (portions where noise was reduced being indicated by arrows in Fig. 5(A)). Quantitatively, the use of the present invention improved the segmental SNR by about 1.7 dB and the cepstrum distance (a spectral envelope distortion) by around 0.2 dB. Besides, an unpleasant impulse-like crackling noise was also reduced.

Moreover, the throughput by the present invention is as small as 0.11 MIPS, which is about 2.5% of the throughput (about 4.4 MIPS) of the ADPCM decoder.

While in the above the present invention has been described as being applied to the ADPCM speech coder/decoder, the invention is also applicable to speech coder/decoders of other systems.

As described above in detail, the incorporation of the comparator-attenuator in the speech coder/decoder makes it possible to decrease an impulse-like noise, which is caused by transmission errors and to improve the reproduced speech quality by about 21% in terms of segmental SNR and about 14% in terms of cepstrum distance. This permits communications of higher quality in a system which includes a radio section in which transmission errors are likely to occur. In addition, since the throughput is very small, the addition of the comparator-attenuator does not markedly increase power consumption. Hence, the present invention is of great utility when employed in radio equipment which is required to be low in power consumption.

Claims (7)

CLAIN1S:

1. A speech coding transmission system comprising: a coder receptive of an input speech signal for producing a coded output of adaptive differential PCM by quantizing a difference between a predictive value from a first composite adaptive predictor and the input speech signal by the use of an adaptive quantizer in accordance with the level of the difference signal to send out the coded output to transmission medium, and a decoder receptive of the coded output from the transmission medium for demodulating the coded output by inversely quantizing the coded output with an inverse adaptive quantizer to produce a decoded residual and by adding the decoded residual to a predictive signal from a second composite adaptive predictor' to reproduce a reproduced speech signal, characterized in that at least one of said coder and said decoder comprises a comparator-attenuator for attenuating the amplitude of an input signal to the first composite adaptive predictor or the second composite adaptive predictor to a value smaller than a predetermined threshold value.

2. A speech coding transmission system according to claim l, in which the first composite adaptive predictor quantizes in accordance with a first quantizer-scale factor from a first quantizer-scale factor adapter; the inverse adaptive quantizer inversely quantizes the coded output in accordance with a second quantizer-scale factor from a second quantizer-scale factor adaptor, and the input signal to the first composite adaptive predictor or the second composite adaptive predictor is attenuated in accordance with a corresponding one of the first quantizer-scale factor and the second quantizer-scale factor.

3. A coder comprising: input terminal means receptive of an input speech signal; a composite adaptive predictor for producing a predictive signal; a subtractor for obtaining a difference between the input speech signal and a predictive signal to produce an input residual signal; an adaptive quantizer for quantizing the input residual signal in accordance with a quantizer-scale factor to produce a coded output transmitted to transmission medium; a quantizer-scale factor adapter for generating the quantizer-scale factor from the coded output; an inverse adaptive quantizer for inversely quantizing the coded output to produce a reproduced residual applied to the composite adaptive predictor; an adder for adding the reproduced residual on the predictive signal to produce a reproduced signal; and a comparator-attenuator for attenuating the reproduced signal, when the level of the reproduced signal exceeds a threshold value, to produce an attenuated reproduced signal applied to the composite adaptive predictor.

4. A decoder comprising: input terminal means receptive of a coded output from transmission medium; a quantizer-scale factor adaptor receptive of the coded output for producing a quantizer-scale factor; an inverse adaptive quantizer for inversely quantizing the coded output in accordance with the quantizer-scale factor to produce a coded residual; a composite adaptive predictor for producing a predictive signal from the decoded residual.

an adder for adding the decoded residual and the predictive signal to produce a reproduced speech signal; and a comparator-attenuator for attenuating the reproduced signal, when the level of the reproduced speech signal exceeds a threshold value, to produce an attenuated reproduced signal applied to the composite adaptive predictor.

5. A speech coding transmission system substantially as herein described with reference to Figure 1 with or without reference to any of Figures 2, 3 and 5 of the accompanying drawings.

6. A coder substantially as herein described with reference to Figure 1 with or without reference to any of Figures 2, 3 and 5 of the accompanying drawings.

7. A decoder substantially as herein described with reference to Figure 1 with or without reference to any of Figures 2, 3 and 5 of the accompanying drawings.