JP2003099096A - Audio decoding processor and error compensating device used in the processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the occurrence of abnormal sound even though a miss occurs in error detection. SOLUTION: An audio decoding section 21 is used to decode spectrum data in a frame unit from audio coded data. In an error compensation section 23, a frame in which an error is detected by an error compensation section 23 and frames located in front of and the back of the frame are recognized as erroneous frames, an interpolating process is conducted for the error frames employing the data of a normal frame. The spectrum data outputted from the section 23 are successively converted from frequency region data to time region data and decoded audio data are outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio decoding processing apparatus which decodes audio coded data transmitted by radio, for example, and in particular, compensates data of an erroneous frame to suppress deterioration of quality of audio output. Regarding error compensation technology.

[0002]

2. Description of the Related Art In recent years, the practical application of digital mobile communication systems has been rapidly advanced. In such a mobile communication system, in an audio decoding processing device for decoding audio encoded data transmitted by radio,
When a code error occurs on the transmission path, an error is detected by an error detection function such as CRC, and an error process is performed on the detected frame.

As an example of transmission error processing of audio coded data in a conventional mobile communication terminal, FIG. 7 shows a device configuration by a parameter interpolation method.

In FIG. 7, the received audio coded data is first sent to the parameter interpolating device 11 and the error detecting device 12. The error detection device 12 sends error detection information to the parameter interpolation device 11 when an error is detected. The parameter interpolating device 11 selects a frame in which an error has occurred based on the error detection information from the error detecting device 12 and replaces the data of the frame with the parameter of the immediately preceding frame, thereby performing the parameter interpolation process. To do.

In this way, the audio coded data in which the parameters of the error frame are interpolated are sent to the audio decoding processing device 13. Audio decoding processing device 1
3, the encoded data of the frame in which no error is detected is subjected to normal decoding processing, and the encoded data of the frame in which an error is detected and parameter interpolation processing is performed A process for reducing the power of the decoded output according to the number is performed so that the abnormal noise of the decoded output is inconspicuous.

[0006]

However, in the conventional audio decoding processing device as described above, when the error cannot be detected, the transmitted encoded data is decoded, so that sometimes an abnormal noise occurs. Could occur. For a service such as a digital mobile communication system, it is fatal as a service that an abnormal sound is generated when an error occurs, and therefore a method for avoiding such a situation is required.

Here, in a system having an error correction function such as Reed-Solomon in a transmission path to an audio decoding processing device, such as a digital mobile communication system, an error can be corrected if the transmission path error rate is small. it can. At this time, in the audio decoding processing device, the error has already been corrected at the stage of sending the data, and the data can be decoded as error-free data. On the other hand, if the error rate becomes high, the error cannot be corrected or detected, and the error is directly sent to the audio decoding processing device. In this case,
Burst errors will be sent to the audio decoding device. Therefore, in the above system, if the data sent to the audio decoding processing device is erroneous, there is a high possibility of erroneous burst-like data.

On the other hand, in the audio decoding processing device, as an error detection function for input coded data, for example, CR
It has a detection function like C. However, it is difficult to say that the detection capability is perfect, and there are many cases in which an error cannot be detected by CRC. For this reason, there is a problem that erroneous data may be decoded to generate abnormal noise.

The present invention has been made in order to solve the above-mentioned problems. Even if an error detection error occurs due to a burst error or the like occurring in input audio coded data, an abnormal sound is generated. The present invention provides an audio decoding processing device capable of suppressing the above-mentioned error, and suppressing the auditory deterioration of the decoded output even when the error rate is high, and an error compensating device for performing error compensation processing using this device. The purpose is to

[0010]

In order to achieve the above object, an audio decoding processing device according to the present invention comprises an audio decoding unit for inputting the audio coded data and decoding spectrum data in frame units. An error detection unit that inputs audio encoded data and detects an error in the data, and a spectrum data decoded by the audio decoding unit is input to store the spectrum data up to a predetermined frame before the past, and the error detection unit. The frame in which an error is detected and the frames before and after it are regarded as error frames, the spectrum data of each error frame is compensated and output, and the spectrum data of other frames are output as they are. The spectrum data output from the compensator is sequentially converted from the frequency domain to the time domain data and output. Characterized by comprising a frequency / time conversion unit for.

Further, the error compensating apparatus used in the audio decoding processing apparatus according to the present invention is based on a frame memory section for storing the decoded spectrum data in frame units up to a predetermined frame in the past and an error detection result of each frame. Then, the frame in which an error is detected and the frames before and after the frame are regarded as error frames, and the spectrum data of each error frame regarded as an error frame by this error frame selection unit is interpolated and output. And a frame interpolation processing unit that outputs the spectrum data of other frames as they are.

That is, in a system in which errors occur in bursts, there is a high possibility that many errors exist in the data near the frame in which the error is detected. Therefore, the audio decoding processing apparatus and the error compensating apparatus according to the present invention forcibly process several frames before and after the frame in which an error is detected as an error frame, thereby reducing the possibility of abnormal noise due to an error detection error. To do.

Further, according to the present invention, the error rate Y in a certain section is obtained, and when the error rate Y exceeds a predetermined error rate X, a noise component is added to the section or the power of the section is increased. Attenuate. As a result, even if the mute state continues for a long period of time due to the interpolation process, the auditory deterioration of the output signal after error compensation is suppressed.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below in detail with reference to the drawings.

(First Embodiment) FIG. 1 is a block diagram showing the configuration of an audio decoding processing apparatus according to the present invention. In FIG. 1, input audio coded data is obtained by converting the spectrum data of an audio stream from time domain data to frequency domain data on the transmission side,
The data is encoded with a predetermined frame structure and is demodulated and detected by a reception processing system (not shown). This input encoded data is processed by the audio decoding unit 21 and the error detection unit 22.
Sent to.

The error detector 22 determines whether or not the input coded data has an error by, for example, a CRC check, and if there is an error, sends the error detection information to the audio decoder 21 and the error compensator 23.

The audio decoding unit 21 decodes the spectrum data from the input coded data in frame units, and the decoded spectrum data is sent to the error compensating unit 23. However, if there is an erroneous frame according to the error detection information from the error detection unit 22, the spectrum data of the frame is invalidated (all 0) and output.

The error compensator 23 is specifically constructed as shown in FIG. That is, the decoded spectrum data from the audio decoding unit 21 is temporarily stored in the frame buffer memory 231 for the specified number of frames,
The data is sequentially sent to the frame interpolation processing unit 232. On the other hand, the error detection information from the error detection unit 22 is input to the error frame selection unit 233. This error frame selection unit 233
Identifies an erroneous frame from the error detection information, further forcibly regards the frames before and after it as an error frame, selects an error frame, and invalidates the data of the error frame in the memory 231 (all 0s). . The error frame information selected here is used as the frame interpolation processing unit 23.
2 and the output processing unit 234.

The frame interpolation processing section 232 performs the error frame interpolation processing by replacing the frame recognized as an error frame with the data of the latest frame which was determined to be normal in the past. In addition, the output processing unit 234
Is the output data of the frame interpolation processing unit 232.
A process of adding a noise component to the section of the error frame or attenuating the power of the section is performed.

The processing output of the error compensator 23 is sent to the frequency / time converter 24, converted from the frequency domain to the time domain data, and output as decoded audio data.

The operation of the audio decoding processing device configured as described above will be described below. In the present embodiment, the number of adjacent frames N forcibly performing error processing when an error occurs is set to 1, and the number of consecutive frames β that can be interpolated
However, the values of N and β may be set to any other values.

As a system in which the present invention is effective, for example, in a transmission path to an audio decoding processing device,
It is a system with error correction function like Reed Solomon. In such a system, if there are few errors in the transmission path, the error is corrected. Therefore, it has already been corrected at the stage when the data is sent to the audio decoding processing device. Therefore, the data can be decoded as error-free data. However, when the error rate becomes high, the error cannot be corrected and detected, and the error is directly sent to the audio decoding processing device. In this case, burst errors are sent to the audio decoding processing device. Therefore, in the above system, if the data sent to the audio decoding processing device is erroneous, there is a high possibility of erroneous burst-like data. The present embodiment effectively exhibits its action in the system as described above.

First, the encoded input data is input to the audio decoding unit 21 and the error detection unit 22. When an error is detected by the error detection unit 22 in the current input frame, the spectrum data of the frame is not output from the audio decoding unit 21, and the frame buffer memory of the error compensation unit 23 (here, (β + 2 All 0s are stored in 231) (more than one frame is required). At this time, it is determined that there is a high possibility that there is an error even one frame before the error frame to which the error detection information is sent, the corresponding frame in the memory 231 is regarded as an error frame, and the data of that frame is all 0s. Can be rewritten as

In the memory 231, the spectrum data before the current frame by (β + 1) frames is read. If the frame of this data is not an error frame, the frame interpolation processing unit 232 and the output processing unit 2 as they are.
The signal is output via the circuit 34, converted into a signal in the time domain by the frequency / time conversion unit 24, and output. Here, if a large capacity of the memory 231 of the error compensating unit 23 is set, it is possible to output with a delay longer than (β + 1).

That is, when no error is detected by the error detection unit 22, the audio decoding unit 21 performs a decoding process. The decoded spectrum data is input to the error compensating unit 23 and stored once. Here, when an error is detected one frame before, regardless of the contents of the spectrum data of the frame without the error,
The error compensating unit 23 overwrites all with 0.

On the other hand, an error is detected 2 frames before, and 1
When no error is detected before the frame, the frame interpolation processing unit 232 performs interpolation processing. In the interpolation processing, first, the number N of consecutive error frames is counted, and if the number of consecutively interpolable frames β or less, the interpolation processing is performed by the spectrum data of the frames before and after no error. As an interpolation method, a method of taking an average or a method of taking an average by weighting the spectrum data of the preceding and following frames having no error can be considered.

Then, in the output processing unit 234, the spectrum data of the frame interpolated by the preceding and following error-free frames are gradually attenuated in power as they are separated from the preceding and following error-free frames. On the other hand, when the number N of consecutive error frames is larger than the number β of consecutively interpolable frames, all 0s are written in the error frame and the mute process is performed. Below, the number of consecutive error frames N
Will be specifically described by dividing into 1 frame, 2 frames, and 3 frames.

(When the Number N of Continuous Error Frames is 1) The case where the number N of continuous error frames is 1 will be described with reference to FIG.

First, assume that error detection information is sent from the error detection unit 22 only to the (N + 3) th frame as shown in FIG. The error compensator 23 forcibly sets the (N + 2) th frame and the (N + 4) th frame as error frames and writes all 0s, as shown in FIG. 3B. Then, the number N of consecutive error frames is counted, and if it is equal to or less than the number β of consecutively interpolable frames, interpolation processing is performed.

Here, since the number of consecutive error frames N = 3 and the number of consecutively interpolable frames β = 4, there are no error frames before and after ((N + 1) th frame and (N + 5) th frame as shown in FIG. 3C). Eye, the (N + 2) th frame, the (N + 3) th frame, and the (N + 4) th frame are interpolated.

Finally, as shown in FIG. 3D, power is attenuated for the interpolated frame. As a method of attenuating the power, as shown in FIG. 3D, a process of gradually attenuating the error frame from the beginning and returning the power toward the frame that recovers from the error is performed. As a method of muting, the power may be reduced stepwise as described above, or the power may be reduced at a constant rate.

(When the Number N of Continuous Error Frames is 2) The case where the number N of continuous error frames is 2 will be described with reference to FIG.

First, assume that error detection information is sent from the error detection unit 22 to the (N + 2) th frame and the (N + 3) th frame as shown in FIG. 4A. In the error compensator 23, as shown in FIG. 4B, the (N + 1) th frame and (N)
+4) All 0s are written by forcibly setting the frame frame as an error frame. Then, the number N of consecutive error frames is counted, and if it is equal to or less than the number β of consecutively interpolable frames, interpolation processing is performed.

Here, since the number of consecutive error frames N = 4 and the number of consecutive interpolable frames β = 4, there are no error frames before and after (N frame and (N +) as shown in FIG. 4C.
5) frame), the (N + 1) th frame,
The (N + 2) th frame, the (N + 3) th frame, and the (N + 4) th frame are interpolated.

Finally, as shown in FIG. 4D, power is attenuated for the interpolated frame. As a method of attenuating the power, as shown in FIG. 4D, a process of gradually attenuating the error frame from the beginning and returning the power toward the frame that recovers from the error is performed. As a method of muting, the power may be reduced stepwise as described above, or the power may be reduced at a constant rate.

(When the number N of consecutive error frames is 3) The case where the number N of consecutive error frames is 3 will be described with reference to FIG.

First, assume that error detection information is sent from the error detection unit 22 to the (N + 2) th frame, the (N + 3) th frame, and the (N + 4) th frame as shown in FIG. 5A. In the error compensator 23, as shown in FIG.
All 0s are written by forcibly making the (N + 1) th frame and the (N + 5) th frame the error frames.

In this case, the number of consecutive error frames N = 5, which exceeds the number of consecutively interpolable frames β = 4.
As in (c), the interpolation process is not performed, but the mute process is performed as shown in FIG. Note that the mute method may be a method in which the power is gradually reduced in the frames before and after there is no error.

As is clear from the above specific examples, the first
According to the configuration of the embodiment of the present invention, even if there is a frame in which an error is not detected, the frames before and after the frame in which the error is detected are interpolated as error frames and output control is performed. It is possible to suppress sound generation.

(Second Embodiment) Although the generation of abnormal noise can be suppressed by the method described in the first embodiment,
Since the frames before and after the error detection frame are also forcibly processed as error frames, the mute section inevitably becomes long. Here, as a second embodiment of the present invention, a method for suppressing the deterioration of the auditory characteristic when the mute section becomes long will be described with reference to the drawings.

In the section of the time length Z shown in FIG.
When the error detection information is sent from the error detection unit 22 as shown in (a), the error compensation unit 23 forcibly processes the preceding and succeeding frames as error frames as shown in FIG. 6 (b). Here, the description will be made assuming that the number of adjacent frames N forcibly performing error processing when an error occurs and the number of consecutive frames β that can be interpolated are β = 3, but the values of N and β are any other values. May be specified.

In the error compensator 23, interpolation or mute processing is performed, and a power output as shown in FIG. 6C is obtained. However, an output voice in which a normal output without error and a mute section alternately appear tend to give a strong sense of discomfort. In order to solve this problem, it is necessary to suppress the transient change of the signal on the time axis.

Therefore, in the present embodiment, the error compensating unit 23
As shown in FIG. 6D, the output processing unit 234 of FIG. 6 intentionally adds analog noise such as white noise to suppress the unpleasant sensation of mute. Although white noise is added here, other noise components may be added. Furthermore, FIG.
As shown in (e), by attenuating the overall power of the section Z, it is possible to suppress a transitional change in the signal between the section Z and the mute section. Further, as shown in FIG. 6F, a similar effect can be obtained by adding a noise component and then attenuating the power.

As described above, according to the second embodiment,
By forcibly processing the frames before and after the frame in which an error is detected as error frames, generation of abnormal noise is suppressed, and noise components are added and power is increased in the section where the error rate becomes high and mute increases. It is possible to suppress the deterioration of the perceptual quality by suppressing the influence of the transitional signal change that occurs between the mute section and the method such as attenuation.

[0045]

As described above, according to the present invention, even if an error detection error occurs due to a burst error or the like that occurs in input audio coded data, the occurrence of abnormal noise can be suppressed. Further, it is possible to provide an audio decoding processing device capable of suppressing the auditory deterioration of the decoded output even if the error rate becomes high, and an error compensating device which is used in this device and performs an error compensation process.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an audio decoding processing device according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a specific configuration of an error compensating unit according to the first embodiment.

FIG. 3 is a diagram illustrating processing when a one-frame error occurs in the first embodiment.

FIG. 4 is a diagram illustrating a process when a two-frame consecutive error occurs in the first embodiment.

FIG. 5 is a diagram illustrating a process when three consecutive frame errors occur in the first embodiment.

FIG. 6 is a diagram illustrating processing when the error rate is high according to the second embodiment of this invention.

FIG. 7 is a block diagram showing the configuration of a decoding processing system for audio encoded data by a conventional parameter interpolation method.

[Explanation of symbols]

11 ... Parameter interpolation device 12 ... Error detection device 13 ... Audio decoding processing device 21 ... Audio decoding unit 22 ... Error detector 23 ... Error compensator 231 ... Frame buffer memory 232 ... Frame interpolation processing unit 233 ... Error frame selection unit 234 ... Output processing unit 24 ... Frequency / time converter

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G10L 3/00 FF term (reference) 5D045 DA20 5J064 AA01 BA16 BB08 BC01 BC11 BD02 5K014 AA01 BA06 EA01 FA06 5K041 AA02 BB01 CC01 FF32 GG13 GG14 GG16 HH45 HH46

Claims (12)

[Claims] 1. An audio decoding processing device for converting spectrum data of an audio stream from time domain data to frequency domain data and decoding audio coded data coded in a predetermined frame structure, wherein the audio coded data is input. An audio decoding unit that decodes spectrum data frame by frame, an error detection unit that inputs the audio encoded data and detects an error in the data, and a spectrum data that is decoded by the audio decoding unit. The spectrum data up to a predetermined frame in the past is stored, the frame in which an error is detected by the error detection unit and the frames before and after the frame are regarded as error frames, and the spectrum data of each error frame is compensated and output. Error compensation that outputs the spectrum data of the frame An audio decoding processing device, comprising: a unit and a frequency / time conversion unit that sequentially converts the spectrum data output from the error compensator into data in the frequency domain and outputs the data. 2. The error compensating unit compensates a frame in which an error has been detected by the error detecting unit for adjacent frames within N (N is a natural number) frames as error frames. The audio decoding processing device according to claim 1. 3. The number of consecutive error frames of the error compensating unit is α (α is a natural number) until immediately before, and when the current frame has no error, α is a predetermined number of consecutively interpolable frames. When β (β is a natural number) or less, perform compensation processing to interpolate the spectrum data of the erroneous frame by the saved spectrum data of the normal frame in the past and the spectrum data of the current error-free frame The audio decoding processing device according to claim 1, wherein: 4. The error compensating unit performs a compensation process of attenuating the gain of the compensated spectrum data of the error frame in steps or at a constant rate.
The audio decoding processing device described. 5. The error compensating section obtains an error rate Y in a certain section from the error detection result of the error detecting section, and when the error rate Y exceeds a predetermined error rate X, the section is calculated. The audio decoding processing device according to claim 1, wherein a compensation process for adding a noise component to is performed. 6. The error compensating section obtains an error rate Y in a certain section from the error detection result of the error detecting section, and when the error rate Y exceeds a predetermined error rate X, the section is calculated. The audio decoding processing device according to claim 1, wherein a compensation process for attenuating the power of is performed. 7. The audio coded data is used in an audio decoding processing device for converting spectrum data of an audio stream from time domain data to frequency domain data and decoding the audio coded data coded with a predetermined frame structure. A frame memory unit for inputting an error detection result in the audio coded data together with the spectrum data decoded in frame units, and storing the decoded spectrum data in frame units up to a predetermined frame before the past; Based on the results, an error frame selection unit that considers the frame in which an error is detected and the frames before and after it as error frames, and performs an interpolation process on the spectrum data of each error frame that is considered as an error frame by this error frame selection unit. Output the spectrum data of other frames Error compensating apparatus of an audio decoding apparatus characterized by comprising a frame interpolation process unit to output it. 8. The error frame selection unit, based on the error detection result, regards a frame in which an error has been detected as being an adjacent error frame within N (N is a natural number) frames. An error compensating device for an audio decoding processing device according to claim 7. 9. The frame interpolation processing unit, until immediately before, the number of consecutive error frames is α (α is a natural number) frames,
When there is no error in the current frame, and when α is equal to or less than the predetermined number of consecutively interpolable frames β (β is a natural number), the spectrum of the past normal frame stored in the frame memory unit. 8. The error compensating apparatus for an audio decoding processing apparatus according to claim 7, wherein the spectrum data of the erroneous frame is interpolated by the data and the spectrum data of the current error-free frame. 10. The audio decoding process according to claim 7, further comprising an output processing unit that attenuates the gain of the spectral data of the error frame interpolated by the frame interpolation processing unit stepwise or at a constant rate. Device error compensation device. 11. The output processing unit obtains an error rate Y in a certain section, and when the error rate Y exceeds a predetermined error rate X, adds a noise component to the section. An error compensating apparatus for an audio decoding processing apparatus according to claim 10. 12. The output processing unit obtains an error rate Y in a certain section, and when the error rate Y exceeds a predetermined error rate X, attenuates the power in the section. An error compensating apparatus for an audio decoding processing apparatus according to claim 10.