JPH0429415A - Sound encoding/decoding device - Google Patents

Abstract

PURPOSE:To suppress the increment of the amount of transmission data by deciding the possibility of the error of decode data based on metric information when performing Viterbi decoding, outputting a frame interpolation request signal when the possibility of the error is high, and performing the interpolation of a frame at a sound decoder part. CONSTITUTION:Sound 30, after being sound-encoded by a sound encoder part 21, is convolution-encoded by a convolution encoder part 22, and is outputted as the transmission data 33. Reception data 34, after being demodulated by a demodulation part 24, is Viterbi decoded by a Viterbi decoder part 25, and when it is decided that the error is generated in Viterbi decode data at an interpolation decision part 26, the frame interpolation request signal 37 is outputted to the Viterbi decoder part 25 and a sound decoder part 37, and the interpolation of the frame is performed at a sound decoder part 27 by interpolation algorithm decided in advance without using the reception sound encode data 38 of the frame, then, sound 39 is outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a voice code/decoder used in digital communications.

Conventional technology FIG. 2 shows the configuration of a conventional speech code/decoder.

Note that the transmitting section is shown in the upper part of the figure, and the receiving part is shown in the lower part.

In the transmitting section, 1 is an audio encoding section, which divides the audio 10 into several tens of 118 frames, encodes the audio, and outputs audio encoded data 11. 2 is a parity adding section which adds parity to the encoded audio data 11 and outputs the encoded audio data 12 with parity. 3 is a convolutional encoding unit, which convolutionally encodes the audio encoded data 12 with parity and outputs convolutionally encoded data 13; 4 is a modulation unit, and convolutional encoded data 1
3 is modulated and transmitted as transmission data 14 to the receiving section.

In the receiving section, 5 is a demodulating section, which demodulates the received data 15 received from the transmitting section and outputs demodulated data 16. 6 is a Viterbi decoding unit, which decodes the demodulated data 16 using the Viterbi algorithm and generates Viterbi encoded data 17.
Output. 7 is an error detection unit which checks the parity of the Viterbi decoded data 17; If this Viterbi decoded data 17 is correct, it is output as received audio encoded data 18, and if an error is detected in the Viterbi decoded data 17, a frame interpolation request signal 19 is output. Reference numeral 8 denotes an audio decoding unit, which decodes the received audio encoded data 1 when the error detection unit 7 does not output the frame interpolation request signal 19.
8 is decoded and audio 20 is output, and when the frame interpolation request signal 19 is output in the error detection unit 7, that frame is interpolated by a preset interpolation algorithm without using the received audio encoded data 18 of that frame. is performed and the audio 20 is output.

As shown in 2, in the conventional audio encoding/decoding device, the transmitting section divides the audio 10 into frames and adds parity to the audio encoded data 11, convolutionally encodes this and transmits it, and the receiving section After demodulating the received data 15, the received data 15 is Viterbi-decoded, and when the error detection unit 7 detects an error in the Viterbi-decoded data 17, a frame interpolation request signal 19 is output, so that the audio decoding unit 8 performs frame interpolation. This prevents the output of aurally unbearable sound due to incorrect data decoding.

Problems to be Solved by the Invention However, as described above, in the conventional audio encoding/decoding device, parity is added to the audio encoded data 11, and the audio encoded data 12 with parity is convolutionally encoded and transmitted. , the amount of transmitted data 14 is large (
There is a problem with becoming.

The present invention solves these conventional problems,
By not adding parity to audio encoded data, it is transmitted while suppressing the increase in data amount, and if there is an error in the received audio encoded data at the receiver, the frame can be interpolated and output as audio. The purpose of the present invention is to provide a speech code/decoding device.

Means for Solving the Problems In order to achieve the above-mentioned object, the present invention includes a transmitting section that includes a voice encoding section that divides the voice into frames and encodes the voice, and a voice encoding section that folds the voice encoded data from the voice encoding section. The receiver includes a convolutional encoder that performs convolutional encoding, a modulator that modulates the convolutionally encoded data from the convolutional encoder and transmits the E-modulated data, and a demodulator that demodulates the received data from the transmitter. a Viterbi decoding unit that Viterbi decodes the demodulated data from the demodulation unit, an audio decoding unit that decodes the Viterbi decoded data from the Viterbi decoding unit, and decoding based on metric information at the time of Viterbi decoding. The apparatus includes a frame interpolation determining section that determines the possibility that the data is erroneous and outputs a frame interpolation request signal to the audio decoding section when the possibility is high.

Effects The present invention has the following effects due to the above-described configuration. In the transmitter, the voice is encoded and convolutionally encoded, and then modulated and transmitted. That is, the data is transmitted without adding parity to the data. In the receiving section, after demodulating the received data, Viterbi decoding is performed, and the possibility of error in the decoded data is determined based on the metric information during the second Viterbi decoding, and when the possibility of error is high, a frame interpolation request signal is sent. The data is output and the frame is interpolated by the audio decoding unit. This suppresses an increase in the amount of transmitted data.

Embodiment FIG. 1 shows the configuration of an embodiment of the present invention. In addition, also in this example, the transmitter is shown in the upper part of the figure.
The receiving section is shown at the bottom.

In the transmitting section, 21 is an audio encoding section, which divides the audio input 30 into frames of several tens of AS, encodes the audio, and outputs audio encoded data 31. 22 is a convolutional encoding unit that convolutionally encodes audio encoded data 31 and outputs convolutionally encoded data 32. 23 is a modulation section that demodulates the convolutionally encoded data 32 and outputs it as transmission data 33 to the reception section.

In the receiving section, 24 is a demodulating section, which demodulates the received data 34 received from the transmitting section and outputs demodulated data 35. 25 is a Viterbi decoding unit which decodes the demodulated data 35 using the Viterbi algorithm, stores branch metrics in each state at each time point during decoding, and when decoding demodulated data at a certain time point by traceback. The branch metric in the selected state at that point is output as metric information 36. 26 is a frame interpolation determination unit which, upon receiving the metric information 36 from the Viterbi decoding unit 25, calculates the sum of past branch metrics from that point up to the constraint length of the convolutional code;
If the sum exceeds a preset threshold, it is determined that an error has occurred in the Viterbi decoded data, and a frame interpolation request signal 37 is output. Note that the threshold set in the frame interpolation determination unit 26 is determined depending on the type of communication, the speech code decoding, and the relationship between the proportion of errors that cannot be corrected by Viterbi decoding, that is, the residual error rate, and the value of the metric that gives it. It can be set in advance based on the residual error rate allowed by the system. 27 is an audio decoding unit, and the frame interpolation request signal 3 is
7 is not output, the received audio encoded data 38 output from the Viterbi decoder 25 is decoded and audio 39 is output. Further, when the frame interpolation request signal 37 is outputted in the frame interpolation determination unit 26, the received audio encoded data 38 of the erroneous frame is not used, but that frame is interpolated using a predetermined interpolation algorithm, and the audio 39 is output. .

Next, the operation of the above embodiment will be explained.

In the transmitter, the voice 30 is encoded by the voice encoder 21, convolutionally encoded by the convolutional encoder 22, modulated by the modulator 23, and output as transmission data 33. Therefore, parity is not added to the second transmission data 33 as in the conventional example.

In the receiving section, the received data 34 is demodulated by the demodulating section 24, then Viterbi decoded by the Viterbi decoding section 25, and then Viterbi decoded by the Viterbi decoding section 25 in the frame interpolation determining section 26.
If it is determined that there is no error in the Viterbi decoded data based on the metric information 36 received from 5, the Viterbi decoder 25 outputs the decoded data as received audio encoded data 38. This received audio encoded data 38 is encoded by the audio encoder 27 and audio 39 is output. Furthermore, when the interpolation determining section 26 determines that an error has occurred in the Viterbi decoded data, a frame interpolation request signal 37 is output to the Viterbi decoding section 25 and the audio decoding section 37, and the audio decoding section 27 Without using the received audio encoded data 38 of the erroneous frame, that frame is interpolated using a predetermined interpolation algorithm, and audio 39 is output.

2, according to the embodiment, the transmitting unit does not add parity to the transmitted data 33, and the receiving unit determines errors in the received audio encoded data 38 based on metric information during Viterbi decoding. If there is a frame interpolation request signal 37, the frame interpolation is performed by outputting the frame interpolation request signal 37, thereby making it possible to suppress an increase in the amount of transmitted data.

Effects of the Invention As is clear from the embodiments described above, the present invention allows the transmitter to transmit data without adding parity to encoded audio data.
While the received data is convolutionally encoded, modulated, and transmitted, the receiving section demodulates the received data, then Viterbi decodes it, determines errors in the received audio encoded data based on metric information during Viterbi decoding, and detects errors. Since the frame is interpolated when there is a problem, it is possible to suppress an increase in the amount of transmitted data and to prevent the output of audibly intolerable sound due to decoding of erroneous data.

[Brief explanation of drawings]

FIG. 1 is a schematic block diagram of a speech code decoding device according to an embodiment of the present invention, and FIG. 2 is a schematic block diagram of a conventional speech code decoding device. 21... Audio encoder, 22... Convolutional encoder, 23... Modulator, 24... Demodulator, 25...
Viterbi decoding unit, 26... frame interpolation determination unit, 27
...Audio decoding unit, 30...Audio input, 31...
Audio encoded data, 32... Convolutional encoded data, 33... Transmission data, 34... Reception data, 35
... Demodulated data, 36... Metric information, 37.
. . . Frame interpolation request signal, 38 . . . Received audio encoded data, 39 . . . Audio output.

Claims (2)

[Claims] (1) The transmitter includes a voice encoding unit that divides voice into frames and encodes the voice, a convolutional encoder that convolutionally encodes the voice encoded data from the voice encoder, and a convolutional encoder that convolutionally encodes the voice encoded data from the voice encoder. The receiving section includes a demodulating section that demodulates the received data from the transmitting section, and a Viterbi-decoding demodulated data from the demodulating section. a Viterbi decoding unit that decodes the Viterbi decoded data from the Viterbi decoding unit, an audio decoding unit that decodes the Viterbi decoded data from the Viterbi decoding unit, and determines the possibility that the decoded data is incorrect based on metric information at the time of the Viterbi decoding. and a frame interpolation determination unit that outputs a frame interpolation request signal to the audio decoding unit when there is a high possibility of this occurring. (2) The audio code/decoding apparatus according to claim (1), wherein the possibility that the Viterbi decoded data is erroneous is determined using a preset threshold.