JPH05241599A - Voice waveform decoding device - Google Patents

Abstract

PURPOSE:To eliminate the unpleasant feeling of a receiver by providing a minimum code output circuit outputting a minimum code and a selective circuit selecting any one side of an adaptive difference code and the minimum code and inputting it to an inverse quantization circuit. CONSTITUTION:This device is provided with the inverse quantization circuit 1, an adder circuit 2, a predicting circuit 3, and further, the minimum code output circuit 6 outputting the positive minimum code '000' or the negative minimum code '100' corresponding to the positive/negative of the inverse quantization value of the adaptive difference PCM (ADPCM) code of one piece before and the selective circuit 7 switching the inputted ADPCM code and the output of the minimum code output circuit 6 and outputting it to the inverse quantization circuit 1. Then, when an transmission error occurs at a certain time and it is detected, a soundless output instruction signal M instructing so as to output a soundless signal is sent to the selective circuit 7. When the soundless output instruction signal M is inputted to the selective circuit 7, the selective circuit 7 is switched from the input of the ADPCM code till then to the output of the minimum code output circuit 6 and the output of the circuit 6 is inputted to the inverse quantization circuit 1. Then, an output waveform is changed from the border point of the soundless signal output gradually and an unpleasant to listen higher harmonic noise is eliminated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech waveform decoding apparatus, and more particularly to a speech waveform decoding apparatus for speech waveform decoding in a digital transmission type cordless telephone.

[0002]

2. Description of the Related Art In recent years, cordless telephones using an analog transmission system have become widespread, but problems such as limitation of channel frequency allocation and voice leakage have occurred. Therefore, as a method for solving these problems, with the background of the development of the digital signal processing theory and the progress of the high integration technology of IC,
Cordless telephones using digital transmission methods have been developed. Among them, a voice codec that performs voice encoding and decoding in order to reduce the amount of data on the transmission path is being developed. An adaptive differential PCM (ADPCM) system, which is a waveform coding system that is easy to process, is about to be adopted as a voice coding system of a cordless telephone.

By the way, the transmission line of the cordless telephone is a wireless transmission line. Therefore, there are more transmission errors than the wired transmission path. If the voice signal encoded on the transmitting side is erroneous on the transmission path, the voice decoded on the receiving side becomes noise and the content cannot be heard correctly. In addition, the listener is more uncomfortable. Therefore, when a transmission error occurs, the decoded voice is not output but a silent signal is output instead. This prevents the listener from feeling uncomfortable. This is the mute function.

FIG. 4 is a block diagram showing an example of a conventional speech waveform decoding apparatus which is a backward type ADPCM decoding apparatus having a mute function. As shown in FIG. 4, the conventional speech waveform decoding apparatus includes an inverse quantization circuit 1,
The adder circuit 2, the prediction circuit 3, the switch 4, and the output buffer 5 are provided.

As shown in FIG. 5, the inverse quantizer circuit 1 has an A
An inverse quantization table 11 in which an inverse quantization value corresponds to a DPCM code, a coefficient table 12 in which an quantization width coefficient corresponds to an ADPCM code, multiplication circuits 13 and 14, and a quantization width buffer 15 are provided. Was configured.

Table 1 shows an example of the dequantized value and the quantized width coefficient corresponding to the ADPCM code, which are the contents of the dequantized table 11 and the coefficient table 12, respectively.

[0007]

[Table 1]

Next, the operation of the conventional speech waveform decoding apparatus will be described.

As a premise, the ADPC input at time t
Quantization width buffer 15 with M code as binary number "101"
Is the quantization width calculated from the past ADPCM code.
It is assumed that 1'is stored.

First, A when there is no transmission error at time t
Decoding of the DPCM code will be described.

First, the input ADPCM code '10
1 ′ is transferred to the inverse quantization table 11 and the coefficient table 12 in the inverse quantization circuit 1. ADPC transferred
The M code “101” is converted into an inverse quantization value “−2” and a quantization width coefficient “1” in the inverse quantization table 11 and the coefficient table 12, respectively. The output “−2” of the inverse quantization table 11 and the quantization width “1” stored in the quantization width buffer 15 are multiplied by the multiplication circuit 14. On the other hand, the multiplication circuit 13 multiplies the quantization width coefficient “1” by the quantization width “1” stored in the quantization width buffer 15, and the output “1” is returned to the quantization width buffer 15.

Next, the output "-2" of the multiplication circuit 14 is added to the output of the prediction circuit 3 by the addition circuit 2 as a difference value. The addition result becomes a PCM code and is output via the switch 4 and the output buffer 5. The addition result is transferred to the prediction circuit 3 for the next sample.

Next, the silent signal output when a transmission error occurs and is detected at time t + 1 will be described.

In this case, a silence signal output instruction signal M for instructing to output a silence signal is sent to the switch 4, and the switch 4 is cut off so that the output of the adder circuit 2 is not transferred to the output buffer 5. That is, the PCM code immediately before disconnection, that is, the DC waveform outside the audible range is continuously output until the switch 4 is connected next time. A curve C shown in FIG. 2 is an example of an output waveform of the conventional speech waveform decoding device. The waveform of a silent signal when the sample after time t + 1 disappears due to a transmission error is shown. The portion indicated by the dotted line of the curve C is a waveform assuming that the sample has not disappeared. It is shown that the output waveform changes rapidly at time t + 1, which is the boundary point of the silent signal output.

[0015]

The above-mentioned conventional speech waveform decoding apparatus has a drawback in that high-frequency noise is generated due to the abrupt change of the output waveform at the boundary point of the silent signal output.

[0016]

A speech waveform decoding apparatus according to the present invention comprises an inverse quantization circuit for converting an adaptive differential code into an inverse quantized value, and a quantization width coefficient smaller than 1 when the adaptive differential code disappears. Corresponding to the positive / negative sign of the inverse quantized value before erasure, which is the inverse quantized value immediately before the erasure, and the numerical value of the adaptive difference code having a value smaller than the numerical value of the inverse quantized value before erasure. A minimum code output circuit that outputs a minimum code that is
A selection circuit for selecting one of the adaptive difference code and the minimum code and inputting it to the dequantization circuit is configured.

[0017]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram showing a first embodiment of a speech waveform decoding apparatus of the present invention. This embodiment is a backward type AD having a mute function similar to the conventional example.
It is an example of a PCM decoding device.

As shown in FIG. 1, the speech waveform decoding apparatus according to the present embodiment has a dequantizing circuit 1, an adding circuit 2 and a predicting circuit 3 which are the same as those of the conventional example, in addition to the preceding one. Corresponding to the sign of the dequantized value of the ADPCM code, its positive minimum code '0
A minimum code output circuit 6 for outputting 00 'or a negative minimum code' 100 ', and a selection circuit 7 for switching the input ADPCM code and the output of the minimum code output circuit 6 and inputting them to the inverse quantization circuit 1 are provided. Is configured.

As described in the conventional example, the inverse quantization circuit 1 is provided with the inverse quantization table 11, the coefficient table 12, the multiplication circuits 13 and 14, and the quantization width buffer 15. ..

Next, the operation of this embodiment will be described.

Similar to the above-mentioned conventional example, it is premised that the ADPCM code input at time t is a binary number "101", and the quantization width buffer 15 has a quantization width "calculated from a past ADPCM code". It is assumed that 1'is stored.

First, A when there is no transmission error at time t
Decoding of the DPCM code will be described.

First, the input ADPCM code '10
1 ′ is transferred to the inverse quantization table 11 and the coefficient table 12 in the inverse quantization circuit 1 via the selection circuit 7. The transferred ADPCM code “101” is converted into an inverse quantization value “−2” and a quantization width coefficient “1” in the inverse quantization table 11 and the coefficient table 12, respectively.
The output "-2" of the inverse quantization table 11 and the quantization width "1" stored in the quantization width buffer 15 are the multiplication circuit 1
It is multiplied by 4. On the other hand, the multiplication circuit 13 multiplies the quantization width coefficient “1” by the quantization width “1” stored in the quantization width buffer 15, and the output “1” is returned to the quantization width buffer 15.

Next, the output "-2" of the multiplication circuit 14 is added to the output of the prediction circuit 3 by the addition circuit 2 as a difference value. The addition result is output as a PCM code. The addition result is transferred to the prediction circuit 3 for the next sample.

Next, a silent signal output when a transmission error occurs and is detected at time t + 1 will be described.

In this case, a silence output instruction signal M for instructing to output a silence signal is sent to the selection circuit 7. When the silence output instruction signal M is input, the selection circuit 7 switches the input of the ADPCM code so far to the output of the minimum code output circuit 6 and inputs it to the dequantization circuit 1. As described above, the minimum code output circuit 6 outputs the positive minimum code '000' or the negative minimum code '100' corresponding to the positive or negative of the dequantized value of the preceding ADPCM code. .. Time t
The inverse quantized value at is −2 and is negative. The minimum code '100' is transferred to the inverse quantization table 11 and the coefficient table 12 of the inverse quantization circuit 1 via the selection circuit 7. Inverse quantization table 11 and coefficient table 1
In 2, the input minimum code "100" is converted into the inverse quantized value "-1" and the quantized width coefficient "0.8", respectively. The output “−1” of the inverse quantization table 11 and the quantization width “1” stored in the quantization width buffer 15 are multiplied by the multiplication circuit 14. On the other hand, the multiplication circuit 13 multiplies the quantization width coefficient "0.8" by the quantization width "1" stored in the quantization width buffer 15, and outputs "0.
8'is returned to the quantization width buffer 15.

Next, the output "-1" of the multiplication circuit 14 is added to the output of the prediction circuit 3 by the addition circuit 2 as a difference value. The addition result is output as a PCM code. The addition result is transferred to the prediction circuit 3 for the next sample.

The above processing is repeated until the silent output instruction signal M is cut off. A curve A shown in FIG. 2 is an example of an output waveform of the speech waveform decoding apparatus of this embodiment. Time t + 1
The waveforms of silent signals when the following samples are lost due to a transmission error are shown. The figures below the decimal point are rounded off. The portion shown by the dotted line of the curve A is a waveform assuming that the sample has not disappeared. It is shown that the output waveform gradually changes from time t + 1 which is the boundary point of the silent signal output.

Next, a second embodiment of the present invention will be described.

FIG. 3 is a block diagram showing a second embodiment of the speech waveform decoding apparatus of the present invention. This embodiment is different from the first embodiment described above in that this embodiment is an example of a forward-type adaptive DM (ADM) decoding device having a mute function, and accordingly, some constituent elements are It is changing.

As shown in FIG. 3, the speech waveform decoding apparatus according to the present embodiment has an adding circuit 2 and a predicting circuit 3 according to the first embodiment.
In addition to the multiplication circuits 13 and 14 and the quantization width buffer 15, when the input DM code is a binary number "1", it is "1".
, A dequantization table 16 for converting each to "-1" when it is "0", a silence signal buffer 17 for storing the preceding dequantized value, and a value smaller than "1", for example, "0."
A silent signal coefficient output circuit 18 for outputting 6'and selection circuits 19 and 20 are provided.

Next, the operation of this embodiment will be described.

As a premise, it is assumed that the DM code input at time t is a binary number "0" and the quantization width coefficient "1.2" is input. Further, it is assumed that the quantization width buffer 15 stores the quantization width '1.67' calculated from the past DM code.

First, the decoding of the forward ADM code when there is no transmission error at time t will be described.

The inputted quantization width coefficient “1.2” is transferred to the multiplication circuit 13 via the selection circuit 20 and multiplied with the quantization width “1.67” which is the content of the quantization width buffer 15. To be done. The output “2” is returned to the quantization width buffer 15. On the other hand, the input DM code “0” is converted into the inverse quantized value “−1” in the inverse quantization table 16. The output'-1 'is stored in the silence signal buffer 17 and, at the same time, transferred to the multiplication circuit 14 via the selection circuit 19. The inverse quantized value “−1” and the output “2” of the multiplication circuit 13 are multiplied by the multiplication circuit 14. The output “−2” of the multiplication circuit 14 is added to the output of the prediction circuit 3 by the addition circuit 2 as a difference signal. The addition result is output as a PCM code. The addition result is transferred to the prediction circuit 3 for the next sample.

Next, a silent signal output when a transmission error occurs and is detected at time t + 1 will be described.

In this case, the silent output instruction signal M for instructing to output the silent signal is sent to the selection circuits 19 and 20. When the silence output instruction signal M is input, the selection circuit 20 switches the input of the quantization width coefficient up to the output “0.6” of the silence signal coefficient output circuit 18 and inputs it to the multiplication circuit 13. Silence signal coefficient output circuit 18 by multiplication circuit 13
Output of '0.6' and the contents of the quantization width buffer 15 '
It is multiplied by 2 '. The output “1.2” is returned to the quantization width buffer 15. At the same time, it is transferred to the multiplication circuit 14. On the other hand, the selection circuit 19 outputs the output of the dequantization table 16 to the output of the silence signal buffer 17 '.
-1 'is input to the switching multiplication circuit 14. Multiplication circuit 14
Thus, the output "-1" of the silence signal buffer 17 and the output "1.2" of the multiplication circuit 13 are multiplied. The output "1.2" of the multiplication circuit 14 is added to the output of the prediction circuit 3 by the addition circuit 2 as a difference signal. The addition result is output as a PCM code. The addition result is transferred to the prediction circuit 3 for the next sample.

The above processing is repeated until the silent output instruction signal M is cut off. A curve B shown in FIG. 2 is an example of an output waveform of the speech waveform decoding apparatus of this embodiment. Time t + 1
The waveforms of silent signals when the following samples are lost due to a transmission error are shown. The figures below the decimal point are rounded off. The part indicated by the dotted line of the curve B is a waveform assuming that the sample has not disappeared. It is shown that the output waveform gradually changes from time t + 1 which is the boundary point of the silent signal output.

[0040]

As described above, the speech waveform decoding apparatus of the present invention selects the minimum code output circuit for outputting the minimum code, the adaptive differential code or the minimum code, and the inverse quantization circuit. Since the output waveform does not change abruptly from the boundary point of the silent signal output by providing a selection circuit for inputting to the speaker, no unpleasant high-frequency noise is generated and the receiver It has the effect of not giving discomfort to the person.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of a speech waveform decoding apparatus of the present invention.

FIG. 2 is an output waveform diagram showing an example of an operation in the first and second embodiments and the conventional speech waveform decoding device.

FIG. 3 is a block diagram showing a second embodiment of the speech waveform decoding apparatus of the present invention.

FIG. 4 is a block diagram showing an example of a conventional speech waveform decoding device.

FIG. 5 is a block diagram showing an example of an inverse quantization circuit.

[Explanation of symbols]

 1 Inverse Quantization Circuit 2 Addition Circuit 3 Prediction Circuit 4 Switch Circuit 5 Output Buffer 6 Minimum Code Output Circuit 7, 19, 20 Selection Circuit 11, 16 Inverse Quantization Table 12 Coefficient Table 13, 14 Multiplication Circuit 15 Quantization Width Buffer 17 Silence signal buffer 18 Silence signal coefficient output circuit

Claims (2)

[Claims] 1. A dequantization circuit for converting an adaptive differential code into an inverse quantized value, and a quantization width coefficient smaller than 1 when the adaptive differential code disappears, wherein a positive / negative code is an inverse just before the disappearance. A minimum code output circuit that outputs a minimum code that is an adaptive difference code having a value smaller than the numerical value of the pre-erasure inverse quantized value that matches the positive / negative sign of the pre-erasure inverse quantized value that is a quantized value; A speech waveform decoding apparatus comprising: a selection circuit that selects one of the adaptive differential code and the minimum code and inputs the selected code to the inverse quantization circuit. 2. A quantization width storage circuit for storing a first quantization width, which is a quantization width calculated immediately before in time, and a quantum smaller than 1 when the first quantization width coefficient or differential code disappears. A coefficient output circuit that outputs the second quantization width coefficient that is the quantization width coefficient, and the positive / negative sign matches the positive / negative sign of the before-erasure inverse quantized value that is the inverse quantized value immediately before the erasure, and the numerical value is An inverse quantized value output circuit that outputs a second inverse quantized value that is a value smaller than the numerical value of the before-erasure inverse quantized value, the first quantized width coefficient, and the second quantized width coefficient. A first selection circuit for selecting one of the two, a second selection circuit for selecting one of the first dequantized value and the second dequantized value, the first selection The second quantization width is calculated by multiplying the output of the circuit and the first quantization width stored in the quantization width storage circuit. First multiplying circuit and said second voice waveform decoding unit, characterized in that it comprises a second multiplying circuit for multiplying the output of the selection circuit and the second quantization width.