JP2008176110A - Audio signal processing device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an audio signal processing device and method capable of correcting audio signals with less sense of incongruity, even when some signals are lost. <P>SOLUTION: This is an audio signal processor that repeats decoding of the encoded audio signals and has a stream buffer 11 for holding the encoded audio signals; a decoder 12 for decoding the encoded audio signals stored in the stream buffer 11; a missing period detector 13 for detecting the missing period if some audio signals are missing when the decoder 12 decodes the audio signals; a decoded data buffer 14 for holding the decoded audio signals; and a missing period compensation section 15 for correcting the missing period, by extending the time base of the audio signals in front of the missing period at least in the audio signals, currently stored in the decoded data buffer 14. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

The present invention relates to an audio signal processing device and an audio signal processing method for correcting missing audio signals when decoding an encoded audio signal.

A data stream receiving / reproducing apparatus that repeatedly decodes and reproduces an encoded audio signal that is sequentially input generally has an error correction function for correcting an error of the audio signal. However, if the error correction function is not powerful or if an error in the audio signal occurs for a long time, the error correction function may not be able to correct the error in the audio signal. Something that could not be decrypted occurred. As described above, the fact that error correction is impossible and the audio signal cannot be decoded normally is called “missing” of the audio signal, and the period in which the audio signal is missing is called “missing period”.

Conventionally, various techniques have been proposed as a method for correcting such a missing audio signal. For example, as shown in FIG. 10A, audio signals with time stamps t (n-1) and t (n + 1) are normally received, but audio signals with time stamps t (n) are missing. In other words, when the time stamp t (n) is the missing period t, as shown in FIGS. 10 (b) to 10 (d), the missing period t is silenced or white noise or the like is inserted. Or the audio signal t (n-1) before the missing period is repeatedly reproduced in the missing period t.

However, in the method shown in FIG. 10B to FIG. 10D, the sound (including silence) inserted in the missing period is different from the sound that originally exists. Therefore, since there is no connection before and after the missing period, the sound becomes discontinuous, and an unnatural and uncomfortable sound is reproduced. In some cases, when the sound is interrupted at the discontinuous portion, a sound with an uncomfortable feeling of “buzz” may be reproduced.

Therefore, based on the time lag of the time stamp included in the audio data corresponding to the audio signal, the time axis of the audio data after the time stamp where the time lag begins is extended by the time lag to correct the time lag of the audio signal. A technique has been proposed (see, for example, Patent Document 1). In other words, it is assumed that when audio signals are lost, the audio data is shifted by the missing period, and this time shift is eliminated by expanding the time axis. This eliminates the discontinuity of the sound and reduces the sense of incongruity.
JP 2003-289346 A

However, in Patent Document 1, since only the audio data after the detection of the time deviation is subjected to the time axis expansion, the sense of discomfort tends to increase, and the discomfort tends to increase particularly when the missing period is long.

The present invention has been made to solve the above problems, and an object of the present invention is to provide an audio signal processing apparatus and an audio signal processing method for performing correction processing with less sense of incongruity when an audio signal is lost. It is to be.

An audio signal processing apparatus of the present invention is an audio signal processing apparatus that repeatedly decodes encoded audio signals that are sequentially input, and the apparatus stores a first storage unit that stores encoded audio signals; A decoder that decodes the encoded audio signal held in the first storage unit, and a missing that detects a missing period in which the audio signal is missing when the decoder decodes the encoded audio signal Among the decoded audio signals held in the period detector, the second storage that holds the decoded audio signal, and the second storage,
A missing correction unit that corrects the missing period by extending the time axis of both the decoded audio signal before the missing period and the decoded audio signal after the missing period is provided. .

According to the audio signal processing apparatus of the present invention, the missing period is detected, and the time axis of the audio signal before and after the missing period is extended. Thus, since it correct | amends based on the decoded audio signal before and behind a missing period, it can perform a correction process with less sense of incongruity than it correct | amends based on the data only after a missing period.

The audio signal processing apparatus according to the present invention also provides a band limitation that limits a frequency band of an audio signal to be decoded when the decoder decodes an audio signal after a missing period in which the time axis is expanded by the missing correction unit. It is preferable to further have a part.

According to this audio signal processing device, by limiting the frequency band of the audio signal to be decoded, it becomes easy to decode a large amount of audio signals after the missing period at a high speed,
The speed of the decoding process can be increased.

Further, the audio signal processing apparatus of the present invention lowers the sampling frequency of the decoded audio signal after the missing period lower than the sampling frequency of the decoded audio signal before the missing period, and after the missing period. Down-sampling unit that samples the decoded audio signal and transmits the sampled audio signal to the second storage unit, and monauralization that changes the decoded audio signal after the missing period from stereo to mono It is preferable to further include at least one of the parts.

According to this audio signal processing device, the sampling frequency of the decoded audio signal after the missing period in which the time axis is expanded is made lower than the sampling frequency of the decoded audio signal before the missing period. Since the amount of memory used by the second storage unit holding the decoded audio signal can be reduced, a large amount of the decoded audio signal can be stored in the second storage unit. In addition, the memory used in the second storage unit that holds the decoded audio signal by changing the sound of the decoded audio signal after the missing period in which the time axis is expanded by the missing correction unit from stereo to monaural Since the amount can be reduced, a large amount of decoded audio signals can be stored in the second storage unit.

Further, in the audio signal processing device of the present invention, the missing correction unit is connected until a part of the decoded audio signal after the missing period and a part of the decoded audio signal before the missing period overlap each other or both. It is preferable to extend the respective time axes.

According to this audio signal processing apparatus, when the missing period is relatively short, the decoded audio signals before and after the missing period can be expanded until a part of each other overlaps or the both are joined. In this case, smooth correction data can be created by adding overlapping portions by a window function or the like. In addition, smooth correction data can be created by using a window function or the like in the case of joining together. For this reason, the audio signal can be corrected with less discomfort.

In the audio signal processing device of the present invention, it is preferable that the missing correction unit fades out the audio signal when extending the time axis of the decoded audio signal before the missing period.

According to this audio signal processing apparatus, when the missing period is relatively long, there are cases where the audio signals before and after the missing period cannot be connected even if the decoded audio signals before and after the missing period are expanded. In this case, the portion where the audio signal is missing will remain, but by fading out the audio signal before the missing period, the audio signal will not be suddenly interrupted, and the audio signal will be corrected with less discomfort. Can do.

In the audio signal processing device of the present invention, it is preferable that the missing correction unit fades in the audio signal when extending the time axis of the decoded audio signal after the missing period.

According to this audio signal processing apparatus, when the missing period is relatively long, there are cases where the audio signals before and after the missing period cannot be connected even if the decoded audio signals before and after the missing period are expanded. In this case, a portion where the audio signal is missing remains, but by fading in the audio signal after the missing period, the audio signal does not start suddenly, and the audio signal is corrected with less sense of incongruity. be able to.

Further, in the audio signal processing device of the present invention, the missing correction unit has a predetermined ratio between the time axis of the decoded audio signal before the missing period and the time axis of the decoded audio signal after the missing period. Even when decompressed, if a missing part remains in the decoded audio signal without overlapping the decoded audio signals before and after the missing period, it is preferable to silence the missing part.

According to this audio signal processing apparatus, the missing portion between the fade-out and / or fade-in is silenced, so that the audio signal can be corrected with less discomfort even when the missing period is relatively long.

The audio signal processing method of the present invention is an audio signal processing method for repeatedly decoding encoded audio signals that are sequentially input. The method first decodes and encodes the encoded audio signal. When a decoded audio signal is decoded, a missing period in which the audio signal is missing is detected, the decoded audio signal is retained, and among the decoded audio signals, the decoded audio signal before the missing period And the missing period is corrected by extending the time axis for both of the decoded audio signals after the missing period.

According to the audio signal processing method of the present invention, the missing period is detected, and the time axis of the audio signal before the missing period is extended. Thus, since it correct | amends based on the decoded audio signal before and behind a missing period, it can perform a correction process with less sense of incongruity than it correct | amends based on the data only after a missing period.

ADVANTAGE OF THE INVENTION According to this invention, the audio signal processing apparatus and the audio signal processing method which perform the correction | amendment process with little discomfort by the loss of an audio signal can be provided.

Embodiments of the present invention will be described below with reference to the drawings. In the description of the drawings, the same parts are denoted by the same reference numerals.

With reference to FIG. 1, the configuration of an audio signal processing apparatus according to an embodiment of the present invention will be described. An audio signal processing apparatus according to an embodiment of the present invention is a stream-type data processing apparatus that repeatedly decodes encoded audio signals that are sequentially input. The audio signal processing apparatus has a stream buffer (first buffer) that holds an encoded audio signal.
Storage unit) 11, a decoder 12 that decodes the encoded audio signal held in the stream buffer 11, and the audio signal is lost when the decoder 12 decodes the encoded audio signal. And a missing period detector 13 that detects the missing period. Also,
When the decoder 12 decodes the audio signal after the missing period, the audio signal processing device limits the frequency band of the audio signal to be decoded, and the decoded data that holds the decoded audio signal A buffer (second storage unit) 14, a downsampling unit 18 that lowers the sampling frequency of the decoded audio signal and transmits it to the decoded data buffer 14, and changes the decoded audio signal from stereo to monaural to decode data A monauralization unit 17 for transmission to the buffer 14. Furthermore, the audio signal processing device includes an upsampling unit 20 that increases the sampling frequency of the decoded audio signal held in the decoded data buffer 14, and among the decoded audio signals held in the decoded data buffer 14. The decoded audio signal, before the missing period,
And a missing correction unit 15 that corrects the missing period by extending the time axis of both decoded audio signals after the missing period.

The dotted line in FIG. 1 shows the flow of processing of the audio signal when the missing period detector 13 detects the missing period of the audio signal, and the solid line in FIG. 1 shows the case where the missing period detector 13 does not detect the missing period of the audio signal. The flow of audio signal processing is shown.

When the missing period is detected, the missing correction unit 15 corrects the missing period by extending the time axis of the decoded audio signals before and after the missing period.

When the decoder 12 decodes the audio signal after the missing period, the band limiting unit 16
Of the audio signal after the missing period, the frequency band of the audio signal whose time axis is expanded by the missing correction unit 15 is limited.

The downsampling unit 18 includes the decoded audio signal after the missing period.
The sampling frequency of the audio signal whose time axis is expanded by the missing correction unit 15 is made lower than the sampling frequency of the previously decoded audio signal.

The monauralizing unit 17 changes the audio signal whose time axis is expanded by the missing correction unit 15 from the stereo to the monaural among the decoded audio signals after the missing period.

With reference to FIG. 2, the flow of the audio signal processing method using the audio signal processing apparatus of FIG. 1 will be described.

(A) The stream buffer 11 temporarily holds received data such as data read from the data recording medium and data received from wired communication or wireless communication (step S01).
). Received data stored in the stream buffer 11 is sequentially input. Here, “received data” indicates an encoded audio signal.

(B) The decoder 12 repeatedly decodes the received data sequentially input to the stream buffer 11 (step S02). Here, “decoded data” in FIG. 2 indicates a decoded audio signal.

(C) The missing period detector 13 detects the missing period in which the audio signal is missing when the decoder 12 decodes the received data (step S03).

As described above, an audio signal decoder generally has an error correction function for correcting an error in the audio signal. However, if the error correction function is not powerful or if an error in the audio signal has occurred for a long time, the error correction function cannot correct the error in the audio signal, so that the audio signal cannot be decoded correctly.
The missing period detector 13 detects a signal state in which such error correction is impossible and the audio signal cannot be decoded normally as “missing”. For example, if there is an error in 1-bit data in one packet (= 8 bits), it is possible to correct the error because it is possible to correct the error, and it is determined that it is not missing (step S03: NO). If there is an error in the 5-bit data in (8 bits), it is impossible to correct the error and it is determined that the data cannot be decoded and is missing (step S03: YES). The missing period detector 13 also investigates the length of the missing period when missing is detected. This is because the time axis length of the audio signal whose time axis is expanded by the missing correction unit 15 changes according to the length of the missing period.
Further, in FIG. 1, the missing period detection unit 13 is described as one functional unit included in the decoder 12. However, the present invention is not limited to this, and may be configured as a member different from the decoder 12.

(D) When the missing period detector 13 determines that there is a missing period (step S03: YES)
The band limiting unit 16 limits the frequency band of the audio signal whose time axis is expanded by the missing correction unit 15 among the audio signals after the missing period. Thereby, the decoder 1
The processing speed of the decoding process 2 is increased.

(E) The down-sampling unit 18 decodes the sampling frequency of the audio signal whose time axis is expanded by the missing correction unit 15 among the decoded audio signals after the missing period before the missing period. The sampling frequency of the audio signal is set lower (step S05). As a result, the amount of decoded data held by the decoded data buffer 14 is reduced, and the amount of data used by the decoded data buffer 14 can be reduced.

(F) The monauralization unit 17 changes the audio signal whose time axis is expanded by the missing correction unit 15 from the decoded audio signal after the missing period from stereo to monaural (step S06). As a result, the amount of decoded data held by the decoded data buffer 14 is reduced, and the amount of data used by the decoded data buffer 14 can be reduced.

(G) The decoded data buffer 14 holds the audio signal decoded by the decoder 12 (steps S07 and S10). When a missing period is detected (step S0
3: YES) Among the audio signals after the missing period, the audio signal whose time axis is expanded by the missing correction unit 15 is band-limited by the band limiting unit 16 and the sampling frequency is lowered by the down-sampling unit 18. Then, the audio signal changed from stereo to monaural by the monauralization unit 17 is held.

(H) The up-sampling unit 20 samples the audio signal whose sampling frequency is lowered by the down-sampling unit 18 among the decoded audio signals held in the decoded data buffer 14 by increasing the sampling frequency ( Step S
08).

(I) The missing correction unit 15 is the decoded audio signal held in the decoded data buffer 14 and is decoded after the time axis of the decoded audio signal before the missing period and the missing period. The missing period is corrected by extending the time axis of the audio signal (step S09).

FIG. 3A is a graph showing an example of an audio signal lacking the time stamp t (n) along the time axis. When the audio signal processing apparatus of FIG. 1 receives the audio signal of FIG. 3A, the missing period detection unit 13 determines the time stamp t (n) in step S03 of FIG.
It is determined that the audio signal is missing.

FIG. 3B shows the time axis of the audio signal of the time stamp t (n−1) before the missing period t (n) and the time stamp t (n + 1) after the missing period t (n). 5 is a graph showing a state in which the time axis of each audio signal is expanded and the missing period t (n) is corrected. The missing correction unit 15 is shown in FIG.
In step S09, the time axis of the audio signal of the time stamp t (n-1) before the missing period t (n) and the time stamp t (n + 1) after the missing period t (n) Each time axis of the audio signal is expanded to correct the missing period t (n) as shown in FIG.

In the example of FIG. 3, band limitation (step S04), downsampling (step S05), and monauralization (step S06) are performed on an audio signal t (n + 1) after the missing period t (n). Although these processes are not performed, it is preferable to perform these processes. This is because the decoding speed can be improved and the amount of memory used can be reduced. Band limiting (step S
04), downsampling (step S05), monaural processing (step S06), and the like cause deterioration in sound quality, so that these processing may not be performed in consideration of sound quality.

A specific example of the correction processing method according to the length of the missing period will be described with reference to FIGS. Here, unlike the correction processing method of FIG. 3, as shown in FIG. 4, the received data before the missing period is decoded under normal conditions, and the received data or decoded data after the missing period is decoded. Band limitation (step S04), downsampling (step S05), and monauralization (
A case where step S06) is performed will be described.

FIG. 5A is a graph showing a correction processing method when the missing period in FIG. 4 is relatively short. First, an audio signal after the missing period is down-sampled (step S
05), the up-sampling unit 20 converts the audio signal whose sampling frequency is lowered by the down-sampling unit 18 among the decoded audio signals held in the decoded data buffer 14 to the sampling frequency. The so-called up-sampling is performed by increasing the sampling (step S08). Then, the time axis of the decoded data before and after the missing period is expanded. When the missing period is relatively short, the missing correction unit 15 can extend the respective time axes until the decoded data after the missing period and the decoded data before the missing period overlap each other. And FIG. 5 (b)
Add the overlapped parts as shown in. Smooth correction data can be created by applying a window function at the time of addition.

In the example of FIG. 5A, the time axis of the decoded data before and after the missing period is expanded by approximately the missing period, thereby forming a portion that overlaps each other. However, the present invention is not limited to this, and similarly to the example shown in FIG. 3, the time axis of the decoded data before and after the missing period is expanded by approximately half the missing period, Alternatively, the subsequent decoded data may be connected together.

FIG. 6 is a graph showing a correction processing method when the audio signal missing period of FIG. 4 is relatively long. When the missing period is relatively long, unlike in FIG. 5A, the decoded data before and after the missing period are connected even if the time axes of the decoded data before and after the missing period are expanded. Cannot be performed, and a missing portion remains in the decoded data. In such a case, the missing correction unit 15 fades out the decoded data when extending the time axis of the decoded data before the missing period. As a result, the audio signal is not suddenly interrupted, and the audio signal can be corrected with less discomfort. Further, the missing correction unit 15 fades in the decoded data when the time axis of the decoded data after the missing period is expanded. As a result, the audio signal does not start suddenly, and the audio signal can be corrected with less discomfort. When a missing part remains in the decoded data, the missing correction unit 15 makes the missing part silent. By making the missing part between fade-out and / or fade-in silence, it is possible to reduce and reproduce the noise “buzz”.
Therefore, even when the missing period is relatively long, the audio signal can be corrected with less discomfort.
(First embodiment)
With reference to FIG. 7, the configuration of a device that reads the encoded audio signal stored in the memory card 21, decodes it, and outputs it from the speaker 22, which is a first embodiment to which the present invention is applied, will be described. . When the apparatus of FIG. 7 is compared with the audio signal processing apparatus of FIG. 1, the “received data” in FIG. 1 corresponds to the encoded audio signal read from the memory card 21 in FIG. “Data” corresponds to reproduction data output from the speaker 22 of FIG. 1 and 7 are common to the constituent elements.

The encoded audio signal read from the memory card 21 has a sampling frequency of 48 kHz and is a 2-channel stereo signal. Therefore, one frame is 1024 samples × 2 ch = 2048 samples. The stream buffer 11
Stream data for 20 frames can be held, and the decoded data buffer 14
Stream data for 8 frames (= 16384 samples) can be held.

When the missing period detector 13 determines that there is a missing period (step S03 in FIG. 2: YES)
The band limiting unit 16 limits the frequency band of the audio signal whose time axis is expanded by the missing correction unit 15 among the audio signals after the missing period. Thereby, the decoder 1
The processing speed of the decoding process 2 is increased. For example, when the audio signal is mainly composed of a human voice, since the frequency band of the human voice is 4 kHz or less, the frequency band decoded by the decoder 12 is limited to 10 kHz or less, and the frequency band higher than 10 kHz is decoded. do not do. Thereby, the human voice can be sufficiently reproduced.

Further, the down-sampling unit 18 sets the sampling frequency of the decoded data whose time axis is expanded by the missing correction unit 15 among the decoded data after the missing period to 1 of the sampling frequency of the decoded data before the missing period. / 2 is set (step S05 in FIG. 2). As a result, the amount of decoded data is halved. The data amount of one frame is changed from 2048 samples to 1024 samples.

Then, the monauralization unit 17 changes the decoded signal from stereo to monaural (step S06 in FIG. 2). Thereby, since the decoded data is halved from 2ch to 1ch, the data amount of the decoded data is further halved. The data amount of one frame is changed from 1024 samples to 512 samples. Accordingly, since the amount of decoded data held by the decoded data buffer 14 is reduced, more frames can be stored in the decoded data buffer 14. As a result of downsampling and monauralization, the data amount of one frame is reduced to ¼, so that the data use amount of the decoded data buffer 14 can be reduced to ¼.

When the downsampling unit 18 halves the sampling frequency, the sampling frequency is 24 kHz, and the frequency band of the audio signal that can be expressed is 12 kHz or less, but the band limiting unit 16 has already reduced the frequency band to 10 kHz or less during the decoding process. There is no problem because it is restricted.

The upsampling unit 20 samples the audio signal whose sampling frequency has been lowered by the downsampling unit 18 while increasing the sampling frequency (step S08 in FIG. 2). Then, if the missing period is within 4 frames, the missing correction unit 15 expands the time axes of the decoded data for 4 frames before and after the missing period, and adds the overlapped parts using a window function ( Step S09 in FIG.

With reference to FIGS. 8A to 8C, main data processing states in the processing procedure for correcting the audio signal missing period with the apparatus of FIG. 7 will be described. FIG. 8 illustrates an example in which the missing period is 2 frames, and correction is performed by extending from 4 frames before and after the missing period. FIG.
(A) shows a waveform of received data to be decoded. The missing period detector 13 receives the frames 7 and 8.
Is detected as a missing period, the received data of frames 9 to 12 is decoded as a target for time axis expansion. The data reduction unit 19 including the downsampling unit 18 and the monauralization unit 17 halves the data amount of the decoded data. Therefore, as shown in FIG. 8B, the decoded data of the frames 9 to 12 is stored in the decoded data buffer 14 as data for one frame. Since the data capacity of the decoded data buffer 14 is 8 frames, the decoded data of the frames 1 to 12 cannot be stored at the same time by performing normal decoding processing. Therefore, when the data reduction unit 19 reduces the data amount of the decoded data to ¼, the decoded data buffer 14 can store the decoded data of the frames 9 to 12 as one frame.

When decoding up to the 12th frame is completed, generation of reproduction data is started. Frames 1 and 2 perform normal playback. The time axis of 4 frames (frames 3 to 6) before the missing period is expanded by the missing correction unit 15 to 6 frames including the missing period (2 frames). 4 frames after the missing period (frames 9 to 12 in FIG. 8A, frame 7 in FIG. 8B)
) Is restored to 4 frames by the upsampling process, and then the missing correction unit 15
Is further expanded to 6 frames including a missing period (2 frames).

In frames 7 and 8 which are missing periods, decoded data expanded from before and after the missing period are overlapped with each other, so that reproduced data are generated by overlapping them. By applying a window function at the time of overlap, smoother reproduction data can be generated.

When a moving image is shot with a movie camera or the like and recorded on a memory card 21 such as a flash memory and an encoded audio signal is reproduced, reading of the data stream from the memory card 21 fails for some reason, that is, missing Period may occur. Even in such a case, by using the above-described audio signal processing apparatus, it is possible to reliably and easily perform composite processing of audio signals.
(Second embodiment)
The current digital television broadcasting includes a plurality of error correction functions, and it is said that there is almost no possibility of reception data being lost in a device that performs fixed reception such as a home digital television. However, in a system that receives radio waves while moving, such as a portable terminal or car navigation system, it is assumed that radio wave conditions change between buildings or tunnels, and that radio waves are blocked for a long time. Generally, if data of about 1 second or more is lost, error correction cannot be performed, and an error exceeding the error correction range occurs.

In the second embodiment, the sampling frequency is 48 kHz, and one frame is 1024 samples ×
A digital television broadcast receiving apparatus including a decoded data buffer 14 for 10 frames in a 2-channel environment will be described.

FIG. 9A shows a waveform of an audio signal in which the missing period is relatively long and an audio signal of 50 frames (about 1 second) is missing and error correction is impossible. In this case, as shown in FIG. 9B, the missing correction unit 15 doubles the time axis of the audio signal for 10 frames before the missing period to obtain an audio signal for 20 frames. Further, the missing correction unit 15 fades out the audio signal when extending the time axis of the audio signal before the missing period. As a result, the audio signal is not suddenly interrupted, and the audio signal can be corrected with less discomfort.

When the audio signal is restored again after the data loss continues for about 1 second, FIG.
As shown in b), the missing correction unit 15 doubles the time axis of the audio signal for 10 frames after the missing period to obtain an audio signal for 20 frames. Further, the missing correction unit 15 fades in the audio signal when extending the time axis of the audio signal after the missing period. As a result, the audio signal does not start suddenly, and the audio signal can be corrected with less discomfort.

By performing the time axis extension, the missing period is shortened from 50 frames to 30 frames, but a missing part of 30 frames remains in the audio signal after the time axis extension. Therefore, the missing correction unit 15 sets the missing part as a silent section. Fade out or /
In addition, by setting the missing portion between the fade-in as a silent section, it is possible to reduce and reproduce the noise “buzz”. Therefore, even when the missing period is relatively long, the audio signal can be corrected with less discomfort.

  As described above, according to the embodiment of the present invention, the following effects can be obtained.

In an audio signal processing apparatus that repeatedly decodes encoded audio signals that are sequentially input, a stream buffer 11 holds the encoded audio signal, and the decoder 1
2 decodes the encoded audio signal held in the stream buffer 11, and the missing period detector 13 lacks the audio signal when the decoder 12 decodes the encoded audio signal. The period is detected, the decoded data buffer 14 holds the decoded audio signal, and the missing correction unit 15 decodes the decoded audio signal held in the decoded data buffer 14 before the missing period. The missing period is corrected by extending the time axis for both the audio signal and the decoded audio signal after the missing period.

In an audio signal processing method for repeatedly decoding encoded audio signals that are sequentially input, first, the encoded audio signals are decoded. When decoding the encoded audio signal, a missing period in which the audio signal is missing is detected. Holds the decoded audio signal. Of the decoded audio signals, the missing period is corrected by extending the time axis of both the decoded audio signal before the missing period and the decoded audio signal after the missing period.

Conventionally, since the correction was made by extending the time axis of the audio signal after the time stamp at which the time deviation of the time stamp included in the audio signal starts, the sense of incongruity tends to increase. However, in this embodiment, the time axis is extended for audio signals before and after the missing period. As a result, it is possible to perform correction processing with less sense of incongruity than correction based on data only after the missing period. Conventionally, the loss was detected due to the time offset of the time stamp. However, in this embodiment, for example, if there is an error in 1-bit data in 1 packet (= 8 bits), error correction is possible. Therefore, if there is an error in 5-bit data in one packet (= 8 bits), it is impossible to correct the error and it is determined that it cannot be decoded and is missing. To do. Thus, in this embodiment, since the missing period is directly detected, it is possible to accurately detect the missing audio signal.

The audio signal processing apparatus includes a band limiting unit 16 that limits a frequency band of an audio signal to be decoded when the decoder 12 decodes an audio signal after a missing period in which the time axis is expanded by the missing correction unit 15. Also have.

By limiting the frequency band of the audio signal to be decoded, it becomes easy to decode a large amount of audio signals after the missing period at high speed, and the speed of the decoding process can be increased.

The audio signal processing device further includes at least one of a downsampling unit 18 that downsamples and transmits the decoded audio signal to the decoded data buffer 14 or a monaural unit 17 that changes the decoded audio signal from stereo to monaural. Downsampling unit 18
The sampling unit of the decoded audio signal after the missing period in which the time axis is expanded by the missing correction unit 15 is set lower than the sampling frequency of the decoded audio signal before the missing period, and the monaural unit 17 changes the decoded audio signal after the missing period in which the time axis is expanded by the missing correction unit 15 from stereo to monaural.

Preserve the decoded audio signal by lowering the sampling frequency of the decoded audio signal after the missing period where the time axis is stretched lower than the sampling frequency of the decoded audio signal before the missing period Since the amount of memory used by the decoded data buffer 14 can be reduced, a large amount of decoded audio signals can be stored in the decoded data buffer 14. The use of the decoded data buffer 14 that holds the decoded audio signal by changing the sound of the decoded audio signal after the missing period in which the time axis is expanded by the missing correction unit 15 from stereo to monaural. Since the amount of memory can be reduced, a large amount of decoded audio signals can be stored in the decoded data buffer 14.

The missing correction unit 15 extends each time axis until a part of the decoded audio signal after the missing period and a part of the decoded audio signal before the missing period overlap with each other, or both are joined together, Add together.

If the missing period is relatively short, the decoded audio signals before and after the missing period can be stretched until they partially overlap each other. In this case, smooth correction data can be created by adding overlapping portions using a window function or the like, so that the audio signal can be corrected with less discomfort. In addition, smooth correction data can be created by using a window function or the like in the case of joining together. For this reason, the audio signal can be corrected with less discomfort.

The missing correction unit 15 fades out the audio signal when extending the time axis of the decoded audio signal before the missing period.

If the missing period is relatively long, the audio signals before and after the missing period may not be joined even if the decoded audio signals before and after the missing period are expanded. In this case, the portion where the audio signal is missing will remain, but by fading out the audio signal before the missing period, the audio signal will not be suddenly interrupted, and the audio signal will be corrected with less discomfort. Can do.

The missing correction unit 15 fades in the audio signal when extending the time axis of the decoded audio signal after the missing period.

If the missing period is relatively long, the audio signals before and after the missing period may not be joined even if the decoded audio signals before and after the missing period are expanded. In this case, a portion where the audio signal is missing remains, but by fading in the audio signal after the missing period, the audio signal does not start suddenly, and the audio signal is corrected with less sense of incongruity. be able to.

Even if the missing correction unit 15 expands the time axis of the decoded audio signal before the missing period and the time axis of the decoded audio signal after the missing period, the missing part is included in the decoded audio signal. If it remains, the missing part is silent.

By making the missing part between fade-out and / or fade-in silence, the audio signal can be corrected with less discomfort even when the missing period is relatively long.

As described above, the present invention has been described by the embodiment and the first and second examples.
The discussion and drawings that form part of this disclosure should not be construed as limiting the invention. From this disclosure, various alternative embodiments, examples and operational techniques will be apparent to those skilled in the art. That is, it should be understood that the present invention includes various embodiments not described herein. Therefore, the present invention is limited only by the invention specifying matters according to the scope of claims reasonable from this disclosure.

It is a block diagram which shows the structure of the audio signal processing apparatus concerning embodiment of this invention. It is a flowchart which shows the flow of the audio signal processing method using the audio signal processing apparatus of FIG. FIG. 3A is a graph showing an example of an audio signal in which the time stamp t (n) is missing along the time axis, and FIG. 3B is a time stamp before the missing period t (n). The time axis of the audio signal of t (n-1) and the time axis of the audio signal of the time stamp t (n + 1) after the missing period t (n) are respectively expanded, and the missing period t (n) It is a graph which shows the state which correct | amended. Received data before the missing period is decoded under normal conditions, and band limitation (step S04), downsampling (step S05), and monauralization (step S06) are performed on the received data or decoded data after the missing period. It is a figure explaining the Example implemented. FIG. 5A is a graph showing a correction processing method when the missing period of the audio signal in FIG. 4 is relatively short, and FIG. 5B is a part of the overlapped audio signal in FIG. It is a graph which shows the state which added together. 5 is a graph showing a correction processing method when the audio signal missing period of FIG. 4 is relatively long. 1 is a block diagram showing a configuration of a device that reads out an encoded audio signal stored in a memory card 21, decodes it, and outputs it from a speaker 22, which is a first embodiment to which the present invention is applied. FIG. FIGS. 8A to 8C are graphs showing main data processing states in the processing procedure for correcting the audio signal missing period by the apparatus of FIG. FIG. 9A is a graph showing the waveform of the audio signal when the audio signal missing period is relatively long, and FIG. 9B is a graph in which the time axis of the audio signal before and after the missing period is extended and faded out. FIG. 6 is a graph showing a waveform of an audio signal after performing a fade-in process. FIG. 10A to FIG. 10D are diagrams for explaining a conventional method for correcting missing audio data.

Explanation of symbols

11 Stream Buffer 12 Decoder 13 Missing Period Detection Unit 14 Decoded Data Buffer 15 Missing Correction Unit 16 Band Limiting Unit 17 Monaural Unit 18 Downsampling Unit 19 Data Reduction Unit 20 Upsampling Unit 21 Memory Card 22 Speaker

Claims (8)

In an audio signal processing apparatus that repeatedly decodes encoded audio signals that are sequentially input,
A first storage unit for holding the encoded audio signal;
A decoder for decoding the encoded audio signal held in the first storage unit;
A missing period detector that detects a missing period in which the audio signal is missing when the decoder decodes the encoded audio signal;
A second storage unit for holding the decoded audio signal;
Of the decoded audio signals held in the second storage unit, both the decoded audio signal before the missing period and the decoded audio signal after the missing period An audio signal processing apparatus comprising: a missing correction unit that corrects the missing period by extending a time axis. The audio signal processing apparatus according to claim 1, further comprising a band limiting unit that limits a frequency band of an audio signal to be decoded when the decoder decodes an audio signal after the missing period. Sampling the decoded audio signal after the missing period by lowering the sampling frequency of the decoded audio signal after the missing period to be lower than the sampling frequency of the decoded audio signal before the missing period And at least one of a downsampling unit that transmits the sampled audio signal to the second storage unit and a monaural unit that changes the decoded audio signal after the missing period from stereo to monaural. The audio signal processing apparatus according to claim 1, further comprising: The missing correction unit is configured so that each of the decoded audio signal after the missing period and the decoded audio signal before the missing period overlap each other or until both are joined. The audio signal processing device according to claim 1, wherein the audio signal processing device is expanded. The audio signal processing apparatus according to claim 1, wherein the missing correction unit fades out the audio signal when extending a time axis of the decoded audio signal before the missing period. The audio signal processing apparatus according to claim 1, wherein the missing correction unit fades in the audio signal when extending a time axis of the decoded audio signal after the missing period. The missing correction unit may extend the time axis of the decoded audio signal before the missing period and the time axis of the decoded audio signal after the missing period at a predetermined rate. The audio signal processing according to claim 5 or 6, wherein, when the decoded audio signal before and after the missing period does not overlap and the missing part remains in the decoded audio signal, the missing part is silenced. apparatus. In an audio signal processing method for repeatedly decoding encoded audio signals that are sequentially input,
Decoding the encoded audio signal;
Detecting a missing period in which the audio signal is missing when decoding the encoded audio signal;
Holds the decoded audio signal,
Among the decoded audio signals, the missing audio signal is expanded by extending the time axis of both the decoded audio signal before the missing period and the decoded audio signal after the missing period. An audio signal processing method comprising correcting a period.

Images