JP2705201B2 - Adaptive post-filter control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] For example, the present invention relates to an adaptive post-filter control method used for a high-efficiency speech coding system for digital mobile radio systems, and relates to a circuit without sacrificing efficiency by the high-efficiency coding system. In order to reduce the influence of errors, an error detection encoder that generates redundant bits for error detection for the auxiliary information is provided on the encoder side, and whether the received auxiliary information has an error is provided on the decoder side. An error detection decoder for detecting whether or not
A line state monitoring part for monitoring a line state, switching means for performing a switching operation corresponding to an output from the error detection decoder, and a correction processing of a decoded output of the high-efficiency decoder in response to an output of the switching means The encoder side sends main information and auxiliary information to which an error detection redundant bit has been added as audio information, and the decoder side has an error detection decoder which outputs the auxiliary information. When the switching means detects an error, the switching means stops the correction processing, does not detect the error,
If the line state is better than a predetermined state, a correction process for emphasizing the formant information in the decoded output corresponding to the auxiliary information from the high-efficiency decoder is performed. In addition to the correction processing using the auxiliary information, the correction processing is further performed by a predetermined amount corresponding to the line state.

[Industrial applications]

The present invention relates to an adaptive post-filter control method used in, for example, a high-efficiency voice coding scheme for a digital mobile radio system.

In recent years, full-scale introduction of a digital mobile radio system using a high-efficiency voice coding scheme has been considered. However, since the transmission path of a mobile radio system is free space, factors that hinder accurate transmission (for example, Fading), and information often changes in content due to errors during propagation.

Therefore, when error correction is performed by adding a large number of error correction codes to audio information, the transmission band is widened, and transmission efficiency is sacrificed.

For this reason, it is necessary to reduce the effects of line errors without sacrificing the efficiency of the high-efficiency speech coding scheme.

[Conventional technology]

FIG. 6 shows a block diagram of a conventional example. Hereinafter, the operation of the figure will be described.

First, on the encoder side, an analog audio signal is input to the high-efficiency encoder 11. This encoder obtains a linear prediction filter coefficient by taking in an input signal in block units and performing linear prediction analysis, and generates a prediction signal based on the coefficient. Then, the main information, the linear prediction filter coefficient, the pitch period, and the like obtained by encoding the residual signal between the prediction signal and the input signal are output as auxiliary information, for example, 8 kb by the adaptive bit allocation prediction coding (APC-AB) method. The signal is converted into a digital signal of / s and sent to the error correction encoder 12.

The error correction encoder 12 performs an error correction encoding process on the input digital signal to generate redundant bits, and a multiplexing unit 13 adds the redundant bits to the digital signal to form a frame.
The signal is transmitted to the transmission line as a multiplexed signal. The number of redundant bits is, for example, about 1/2 to 1/3 of the number of information bits.

Next, on the decoder side, the input multiplexed signal is demultiplexed by the demultiplexing unit 14 and subjected to error correction decoding by the error correction decoder 15, and then the digital signal in which the error has been corrected is converted to a highly efficient decoder. At step 16, the original audio signal is extracted.

[Problems to be solved by the invention]

Here, when 8 KHz sampling and 8-bit quantization are performed on the audio signal, the coding rate is 64 Kb / s, but the coding rate is 16 Kb / s or 8 Kb / s by performing high-efficiency coding. Are obtained, and the transmission efficiency is improved.

However, for example, when 1/2 redundant bits are added, the efficiency is 12 Kb / s in the case of 8 Kb / s, and 24 Kbs in the case of 16 Kb / s, which does not reflect the efficiency of high-efficiency coding.

The redundancy is proportional to the error correction capability. To increase the error correction capability, a large number of redundant bits are needed. Error correction that is output as it is becomes impossible, and in the worst case, error correction is performed.

In other words, the addition of redundant bits causes a deterioration in efficiency. If the state of the line deteriorates more than expected, there is a problem that correction is impossible and erroneous correction is performed, thereby adversely affecting the line error.

SUMMARY OF THE INVENTION It is an object of the present invention to reduce the effects of line errors without sacrificing the efficiency of the high efficiency coding scheme.

[Means for solving the problem]

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

In the figure, reference numeral 22 denotes an error detection encoder that generates redundant bits for error detection of auxiliary information, 32 denotes an error detection decoder that detects whether or not the received auxiliary information has an error, and 33 denotes a line state. This is the line status monitoring part to be monitored.

Reference numeral 34 denotes switching means for performing a switching operation corresponding to the output from the error detection decoder. Reference numeral 4 denotes an adaptive postfix for correcting the decoded output of the high-efficiency decoder in response to the output of the switching means. It is a filter.

Then, the main information and the auxiliary information to which the redundant bits for error detection are added are transmitted as audio information. When the error detection decoder detects an error in the auxiliary information, the switching means stops the correction processing and detects the error. If the line state is better than a predetermined state, a correction process is performed to emphasize the formant information in the decoded output corresponding to the auxiliary information from the high-efficiency decoder. If there is, in addition to the correction processing using the auxiliary information, the correction processing is further performed by a predetermined amount corresponding to the line state.

[Action]

In general, in error correction, it is necessary to detect an error and return to the original place from the wrong place. However, in the error detection, it is sufficient to detect whether or not the error is wrong, so that the latter has a smaller number of redundant bits than the former.

Therefore, of the main information and the auxiliary information transmitted as voice information from the high-efficiency encoder, a redundant bit for error detection is added to the auxiliary information to monitor an error in the auxiliary information. on the other hand,
The state of the line is monitored on the decoder side, and the correction processing operation of the adaptive post-filter is controlled based on the monitoring state and information on error detection for the auxiliary information.

This is because if there is no error in the auxiliary information, the error of the residual signal can be corrected by passing the decoded output of the high-efficiency decoder through an adaptive post-filter that emphasizes the formant. The effect of line errors can be reduced by further enhancing the formants of the spectral envelope information.

However, if there is an error in the auxiliary information, it is necessary to perform control to turn off the correction processing of this filter in order to perform an erroneous correction processing and emphasize the error.

That is, the error of the main information is estimated by monitoring the error rate of the line to control the correction processing of the adaptive post-filter, and the error detection decoder detects the error of the auxiliary information and turns on / off the correction processing of this filter. I do.

As a result, the effect of line errors can be reduced without sacrificing the efficiency of the high-efficiency coding scheme.

〔Example〕

FIG. 2 is a block diagram of an embodiment of the present invention, FIG. 3 is a diagram showing the configuration of audio information bits, FIG. 4 is a diagram for explaining an adaptive post-filter control, and FIG. 5 is a diagram for explaining the operation of FIG. 5 shows, for example, the waveforms of the same reference numerals in FIG.

Here, the coefficient generator 341 and the switch SW are the switching means 34
, The synthesizer 44, the adder 45, the delay part 41, the coefficient unit 42,
Reference numeral 43 denotes a component of the adaptive post-filter. The same reference numerals indicate the same objects throughout the drawings. Hereinafter, the operation of FIG. 2 will be described with reference to FIG. 3 to FIG.

First, when a speech signal having a spectrum as shown in FIG.
Are output to the multiplexing unit 23, and the auxiliary information is also added to the error detection encoder 22.
Therefore, redundant bits corresponding to this information are generated and added to the multiplexing unit 23, multiplexed by the multiplexing unit to form audio information bits shown in FIG. 3, and transmitted to the decoding side via the line.
The auxiliary information is a linear prediction filter coefficient, a pitch period, and the like in the same manner as described above.

On the decoding side, the high-efficiency decoder 31 decodes the audio information bits input via the demultiplexing unit 35, and a linear prediction filter coefficient (a coefficient indicating the spectral envelope information of the audio and formant information) 311 is used as a coefficient generator. Sent to 311.

The error detection decoder 32 detects the presence or absence of an error in the auxiliary information by using the redundant bits,
Monitors the state of the incoming call level, for example (FIG. 5 (a)
-See).

Then, the operation of the adaptive post-filter is controlled from the presence or absence of an error in the auxiliary information and the line error rate (corresponding to the error rate of the main information) estimated from the incoming level as shown in FIG.

When there is no auxiliary information error and the estimated line error rate is lower than a predetermined value, the error rate of the main information hardly occurs. Therefore, the control signal from the error detection decoder 32 sets the switch SW to the solid line state to generate the coefficient. The formant information is added to the coefficient unit 42 from the unit 311.

Therefore, the output of the high-efficiency decoder passing through the adder 45 is multiplied by a coefficient each time the output is delayed by a predetermined amount in the delay part, and
After being synthesized at 44, the signal is added to an adder 45, and the decoded output is corrected to obtain an audio signal shown in FIG. 5 (a)-.

When the estimated line error rate is higher than a predetermined value, the weighting coefficient (for example, a value in 0 to 1) sent from the line monitoring unit 33 to correct an error in the main information is calculated by a coefficient unit.
In addition to 43, formants are emphasized as shown in FIG. 5 (b)-'.

In other words, when only the coefficient from the coefficient unit 311 is used, the correction filter itself enhances the formant information as shown in (a) in the figure to obtain an audio signal, but when the line error rate is a predetermined value or more, , Noise is heard, and the formant is further emphasized by using a weighting coefficient, as shown in (b), and the noise is reduced. It should be noted that a weighting coefficient is set so as to weaken the emphasis on the formant when the line error rate decreases.

However, if there is an error in the auxiliary information, the adaptive post-filter performs erroneous emphasis as shown by the dotted line in FIG. 5- ", and the sound signal is distorted. It is tilted to the opposite side, and 0 from the coefficient generator 341 is added to the coefficient unit 42. Thus, the adaptive post-filter stops the operation of the correction processing, and the decoded output of the high-efficiency decoder is output as it is.

That is, the adaptive post-filter is controlled by the error detection processing for the auxiliary information on the decoder side and the monitoring of the error rate of the line.
The effect of the line error can be reduced without sacrificing the efficiency by the high efficiency coding method.

〔The invention's effect〕

As described in detail above, according to the present invention, there is an effect that the effect of line errors can be reduced without sacrificing the efficiency of the high efficiency coding scheme.

[Brief description of the drawings]

 FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is a block diagram of an embodiment of the present invention, FIG. 3 is a block diagram of audio information bits, FIG. 4 is an explanatory diagram of adaptive post-filter control, FIG. FIG. 2 is an operation explanatory diagram, and FIG. 6 is a block diagram of a conventional example. In the figure, 21 is a high-efficiency encoder, 22 is an error detection encoder, 31 is a high-efficiency decoder, 32 is an error detection decoder, 33 is a line state monitoring part, 34 is switching means, and 4 is an adaptive post-filter. Show.

Continuing on the front page (72) Inventor Yoshiki Tanaka 1015 Uedanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Prefecture Inside Fujitsu Limited (72) Inventor Shigeyuki Mari 1015 Uedanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited (56) References JP-A-60-173600 (JP, A) JP-A-1-165235 (JP, A) JP-A-62-117423 (JP, A) JP-A-61-108233 (JP, A)

Claims (1)

(57) [Claims] An audio signal for digitally processing an audio signal by a high-efficiency encoder (21), transmitting main information and auxiliary information to a line, decoding the audio signal by a high-efficiency decoder (31), and extracting an audio signal. In the system, an error detection encoder (22) for generating an error detection redundant bit for the auxiliary information is provided on the encoder side, and whether or not the received auxiliary information has an error is detected on the decoder side. An error detection decoder (32), a line state monitoring part (33) for monitoring the line state, and a switching means (34) for performing a switching operation corresponding to an output from the error detection decoder.
And an adaptive post-filter (4) for correcting the decoded output of the high-efficiency decoder in response to the output of the switching means. The added auxiliary information is transmitted as audio information. When the error detection decoder detects an error in the auxiliary information, the switching means stops the correction processing, does not detect the error, and detects the line state. Is better than a predetermined state, a correction process for emphasizing the formant information in the decoded output corresponding to the auxiliary information from the high-efficiency decoder is performed.
An adaptive post-filter control method, wherein if the line state is equal to or less than a predetermined state, a correction processing is further performed by a predetermined amount corresponding to the line state, in addition to the correction processing based on the auxiliary information.