JPH1175157A - Recorder for video signal and audio signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To evade the variance of amplitude and also to use an ATRAC encoder and an ATRAC decoder which are used currently by recording the video and audio signals on a storage medium synchronously with each other for each GOP after compressing both video and audio signals via the MPEG and ATRAC respectively. SOLUTION: The video and audio data are compressed by an MPEG 2 and an ATRAC respectively. At the same time, the sound group position information to which the dummy data are added by the dummy data adding means 80 and 81 are stored in the header information and recorded on a recording medium. Then both video and audio data are recorded sequentially on the recording medium for each GOP. The header information, the compressed video data equivalent to a single GOP and then the audio data equivalent to two channels are recorded in a single GOP period. Thus, the amplitude never causes any variance when the sample number of every sound group is multiplied by a fixed window number and then synthesized.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal and audio signal recording apparatus for recording a video signal and an audio signal compressed in different units in a synchronized manner.

[0002]

2. Description of the Related Art At present, a standardized ATRAC (Adaptive Transform) is used as a method of compressing an audio signal, recording the signal on a recording medium such as a disk or a tape, and reproducing the signal.
form Acoustic Coding) is known. This ATRA
C is adopted in MD (Mini Disk).

The ATRAC will be described with reference to FIGS. FIG. 8 shows an ATRAC encoder for compressing audio data.

In FIG. 8, for example, an audio signal in a frequency band of 0 to 22 kHz is sampled at a sampling frequency of 44.1 kHz, and then input PCM signal obtained by PCM conversion is It is supplied to input terminal 1. This input audio PC
The M signal is divided into a band of 0 to 11 kHz and a band of 11 kHz to 22 kHz by a band dividing filter 2 such as a so-called QMF (Quadrature Mirror Filter) filter. Further, the signal in the band of 0 to 11 kHz is similarly divided into a band of 0 to 5.5 kHz and a band of 5.5 to 11 kHz by a band dividing filter 3 such as a so-called QMF filter.

[0005] 11 kHz from the above-mentioned band division filter 2
The signals in the z to 22 kHz band are supplied to an MDCT (Modified Discrete Cosine Transform) circuit (Modified Discrete Cosine Transformer) (orthogonal transformer) 4 which is an example of an orthogonal switching circuit, and subjected to MDCT processing. The signals in the 5.5 kHz to 11 kHz band from the band division filter 3 are supplied to an MDCT circuit (modified / discrete cosine transform means) (orthogonal transform means) 5 and subjected to MDCT processing. The 0-5.5 kHz band signal from the band dividing filter 3 is supplied to an MDCT circuit (modified / discrete cosine transform means) (orthogonal transform means) 6,
It is processed. In each of the MDCT circuits 4, 5, and 6, an MDCT process is performed based on a block size (information compression parameter) (length of a processing block) determined by a block determining circuit provided for each band.

As described above, means for dividing an input digital signal into a plurality of frequency bands includes, for example, QMF.
There is a filter for this, 1976 RECrochie
re Digital Coding of Speech In Subbands Bell Syst.
Tech.J.Vol.55, No.8 1976. ICAS
SP 83, Boston Polyphase Quadrature Filters-A NewSub
Band Coding Technique Joseph H. Rothweiler describes a filter division method with equal bandwidth. Here, as the above-described orthogonal transform, for example, an input audio signal is divided into blocks in a predetermined unit time (frame), and a fast Fourier transform (FFT), a discrete cosine transform (DCT), a modified DCT transform (MD
There is orthogonal exchange in which the time axis is converted to the frequency axis by performing CT) or the like. ICASSP for MDCT
1987 Subband / Transform Coding Using Filter Bank D
esigns Based On Time Domain Aliasing Cancellation
JPPrincen ABBradley Univ. Of Surrey Royal Melbo
urne Inst. Of Tech.

The orthogonal exchange block size is determined by an orthogonal exchange block size determination circuit, and the result of the determination is supplied to each of the MDCT circuits 4, 5, 6 and the bit allocation calculation circuit 8 and the block of the block is determined. Size information (processing block length information) (information compression parameter)
Is output from the output terminal 10.

The spectrum data on the frequency axis or the MDCT coefficient data (signal components in a two-dimensional block relating to time and frequency) obtained by performing the MDCT processing in each of the MDCT circuits 4, 5, and 6 have a so-called low band. The medium and high frequencies are grouped for each critical band (critical band), and are supplied to the normalization / quantization circuit 7 and the bit allocation calculation circuit 8 by subdividing the critical bandwidth in consideration of the effectiveness of block floating.

This critical band (critical band) is a frequency band divided in consideration of human auditory characteristics, and when a pure tone is masked by narrow band noise of the same strength near the frequency of a certain pure tone. Is the band that the noise has. In this critical band (critical band), the higher the band, the wider the bandwidth, and the above-mentioned 0
The entire frequency band of ２２22 kHz is divided into, for example, 25 critical bands.

The bit allocation calculation circuit 8 takes into account the so-called masking effect and the like based on the block size information and the spectrum data or MDCT coefficient data.
For each of the divided bands in consideration of the above-described critical band and block floating, a masking amount, and an energy or a peak value or the like for each of the divided bands are calculated. It is supplied to a normalization / quantization circuit 7. In the normalization / quantization circuit 7, the above-described block size information (information compression parameter)
(Spectral data or MDC) according to (length of processing block) and the number of bits allocated to each divided band in consideration of the critical band and block floating.
The T coefficient data is requantized (normalized and quantized). The data encoded in this way is taken out via the output terminal 9 in FIG.

In the encoder of FIG. 8, for example, input 16-bit audio data is divided into three bands by two-stage band division filters 2 and 3, and then divided into three bands.
The signals are converted into spectral signals by 4, 5, and 6.

By dividing the band in advance and converting it into a spectrum signal in this way, the spread of the spectrum signal over a wide band is prevented. As a result, for example, even when a signal having a sharp tone characteristic is input to a low frequency region where masking is particularly difficult to perform, the compression can be performed while maintaining a high S / N ratio, so that sufficient sound quality can be guaranteed.

The obtained spectral signal is quantized after being normalized for each coding unit, and the quantized spectral signal itself can be used to calculate how many bits to perform. The above processing is performed for every 512 samples, and the encoded data is recorded.

FIG. 9 shows a decoder for restoring the audio signal thus compressed. The quantized MDCT coefficients of each band, that is, data equivalent to the output signal of the output terminal 9 in FIG. 8 are supplied to the input terminal 11 in FIG. 9 and used block size information (processing block length) (Information compression parameter), that is, data equivalent to the output signal of the output terminal 10 in FIG. 8 is supplied to the input terminal 12 in FIG. Spectrum restoration circuit 13
Then, the bit allocation is released using the adaptive bit allocation information. Next, inverse orthogonal transform (IMDCT) circuits 14, 15, 1
At 6, the signal on the frequency axis is converted to a signal on the time axis. The signals on the time axis of these partial bands are converted by band synthesis filter (IQMF) circuits 17 and 18 in FIG.
The signal is decoded into a full band signal and output to the output terminal 19.

In this decoder, first, a spectrum signal is restored from the recorded data, and then a time series signal of each band is obtained by performing an inverse transform of the MDCT used in the encoder. For example, a 16-bit audio signal is synthesized by the two-stage band synthesis filters 17 and 18.

As a technique for compressing the video signal and the audio signal, recording the compressed signal on a recording medium such as a disk or a tape, and reproducing the compressed signal, a standardized MPEG2 (Moving) method is used.
Picture Experts Group 2) is known. This MP
EG2 is used for DVD (Digital Video Disk), digital broadcasting, and the like.

In this MPEG2, the compression of the video signal is 15
What is performed in units of frames (about 0.5 seconds) is becoming mainstream. The unit of this compression is GOP (Group of Pic
tures). In the NTSC system, the time of this GOP is precisely defined as 1001/2000 seconds. To synchronize the video signal and the audio signal in this unit, the sampling frequency of the audio signal and the GOP frequency (2000 / 1001 HZ) must be in a certain relationship, for example, a relationship of integer ratio.

More specifically, the sampling frequency used in MDs and CDs is 44.1 kHz. At this frequency, the number of samples in one GOP (1001/2000 seconds) is a fractional number of 22,072.05. Therefore, it is difficult to synchronize the video signal and the audio signal. That is, in these standardized compression methods, it is difficult to use them mutually.

In view of the above, ATRAC and MPEG
There has been proposed a video signal and audio signal recording / reproducing apparatus for synchronizing and recording an audio signal and a video signal compressed in two different units.

Next, the previously proposed video signal and audio signal recording / reproducing apparatus will be described with reference to FIGS. In the video / audio signal recording / reproducing apparatus proposed earlier, the sampling frequency of audio data compressed by ATRAC is set to 48.0 kHz used in DVDs and the like. At a sampling frequency of 48.0 kHz, there are 24,024 integer sampling points in one GOP. Thereby,
This makes it possible to synchronize the video signal and the audio signal. The sampling frequency is 48.
The frequency is not limited to 0 kHz, and another frequency, for example, 96.0 kHz may be used.

FIG. 3 is an overall block diagram of an example of the video signal and audio signal recording / reproducing apparatus. First, on the recording side, an audio signal and a video signal are output from a video camera indicated by 21. The audio signal output from the video camera 21 is MIC. As OUT, the L-channel audio signal is supplied to the A / D converter 22, and the R-channel audio signal is supplied to the A / D converter 23. A / D converter 22
And 23 perform digitization by a sampling clock having a frequency of 48.0 kHz supplied from a clock generator (CLK GEN) 25.

The audio data from the A / D converters 22 and 23 is supplied to a memory 26. The memory 26 is for adjusting the timing of video data and audio data. Under the control of the CPU 32, audio data is supplied from the memory 26 to the ATRAC encoder 28. The audio data is, for example, a linear PCM (Pu
lse Code Modulation) data, which is a time-divided L channel and R channel audio signal. ATRAC
In the encoder 28, audio data corresponding to a GOP is divided into 47 sound groups based on a control signal from the CPU 32, as described later. The audio signals divided into the sound groups are supplied to a multiplexer (MUX) 30.

The video signal is sent from the video camera 21 to the CCD
OUT is supplied to the A / D converter 24. A /
The D converter 24 receives 13.5k from the clock generator 25.
A sampling clock having a frequency of Hz is supplied, and the video signal is digitized. The digitized video data is supplied to the camera process circuit 27. In the camera process circuit 27, camera signal processing such as γ correction and aperture correction is performed, and luminance data Y and color difference data C are generated. The color difference data C is composed of time-division color difference data Cb (blue color difference data) and color difference data Cr (red color difference data). Luminance data Y and color difference data C
Are supplied to the MPEG encoder 29 as 8-bit data. In the MPEG encoder 29, C
Based on a control signal such as a signal indicating a GOP from the PU 32, the luminance data Y and the color difference data C
Encoded with EG2. The encoded video data is supplied to the multiplexer 30.

The multiplexer 30 is supplied with the encoded audio data, the encoded video data, and the header information from the CPU 32, and performs addition of the header information and exception processing of the audio signal. The multiplexer 30 outputs recording data in which a video signal, an audio signal, and header information are combined as shown in FIG. The recording data is supplied to the channel coding circuit 31. In the channel coding circuit 31, processes such as (1, 7) RLL modulation, scrambling, and error correction encoding are performed. The data subjected to these processes is recorded on the recording medium 33 as a bit stream. This recording medium 33
Denotes a recordable disk such as a magnetic or optical tape, an MO (magneto-optical) disk, and a semiconductor memory such as a memory card.

Next, on the reproducing side, the data read as a bit stream from the recording medium 33 is supplied to a channel decoding circuit 41. In the channel decoding circuit 41, processes such as demodulation of (1, 7) RLL, descrambling, and error correction are performed. The data subjected to these processes is supplied to the demultiplexer 42. In the demultiplexer 42, header information, video data and audio data are separated from the supplied data, the audio data is supplied to the ATRAC decoder 43, and the video data is supplied to the MPEG decoder 44. The header information separated by the demultiplexer 42 is supplied to the CPU 32. In the CPU 32,
Information necessary for decoding, such as the position of exceptional processing of the audio signal, is detected from the supplied header information, and the information is supplied to the ATRAC decoder 43 and the MPEG decoder 44 as control signals.

In the ATRAC decoder 43, the CPU 32
Is decoded based on the control signal from ATRAC. The decoded audio data is a time-divided audio signal of the L channel and the R channel, and is supplied to the memory 45. The memory 45
Under the control of the CPU 32, the timing of video data and audio data is adjusted. The audio data read from the memory 45 is supplied to a digital filter 47. The digital filter 47 converts the audio data into a D / A converter 48 at a subsequent stage.
, And 49, the L channel audio data is supplied to the D / A converter 48, and the R channel audio data is supplied to the D / A converter 49.
The analogized audio signal is supplied to the audio signal input terminal of the TV monitor 53 via the output terminals 50 and 51, respectively.

The MPEG decoder 44 to which the video data is supplied from the demultiplexer 42 decodes MPEG2 based on a control signal such as a signal indicating a GOP from the CPU 32. Decoded luminance data Y
The color difference data C is supplied to the video encoder 46 as 8-bit data. The decoded color difference data C is time-division color difference data Cr and color difference data Cb. The video encoder 46 converts the supplied luminance data Y and color difference data C into NTSC signals or RGB signals.
The signal is decoded. The video encoder 46 has a D / A
The video encoder 46 outputs an analog NTSC signal or an RGB signal. The NTSC signal or the RGB signal is supplied to the video signal input terminal of the TV monitor 53 via the output terminal 52.

Here, an example of a conventional method of recording video data compressed by MPEG2 and audio data compressed by ATRAC in synchronization will be specifically described. The sampling frequency of the audio signal is set to an integral multiple of the GOP frequency (2000/1001 Hz). Specifically, the least common multiple of the GOP frequency and the sampling frequency is small, and is generally used.
Let 0 kHz be the sampling frequency of the audio signal. The number of samples of audio data in one GOP obtained from this sampling frequency is 24024. ATRAC
The encoder defines eight modes, and in this example, a coding mode: ATRAC is used as an example.
(1), Level1, ChannelNo. = 2 will be described.

In this coding mode, the number of 512 samples of one channel of audio data is encoded into 212 bytes of compressed data. The processing unit of ATRAC with the number of 512 samples is called a sound group (SG). Hereinafter, only the L channel will be described for ease of description. 24024
When the number of samples of the audio data is divided into 512 sample units, the number of samples becomes 46 sound groups + 472 audio data. Therefore, if 47 sound groups are prepared, a fraction is generated, but audio data for one GOP can be stored.

Comparing 24024 samples of audio data for one GOP sampled at the sampling frequency of 48.0 kHz with samples of audio data for one GOP, the number of audio data samples for one GOP is 47 ( SG) × 512 (samples) = 24064 (samples), so a fraction of 40 samples is generated with respect to the number of samples of one GOP.

In order to absorb these 40 samples,
In this example, for 40 sound groups out of the 47 sound groups, the number of samples of audio data actually used is reduced by one to 511, and encoding is executed. Thus, the process of reducing the number of valid samples by one is the exception process. For the remaining seven sound groups, the number of audio data samples is set to 512.
Encoding is performed as the number. The sound groups composed of 512 sound data samples are arranged as evenly as possible in the 47 sound groups. For example, 0, 7, 14, 21, 27, 34, 41
By arranging a sound group consisting of 12 samples of audio data, audio data can be arranged almost uniformly.

40 (SG) × 511 (sample) +7
(SG) × 512 (samples) = 24024 (samples), which is the number of audio data samples for one GOP.
That is, video data and audio data are completed in units of 1 GOP (about 0.5 seconds).

Here, the number of audio data samples is 511.
A technique for encoding the sound groups by ATRAC will be briefly described. As described above, the ATRAC encoding used in this example requires 512 sound data samples per sound group. However, since the number of audio data samples for one GOP is 24024, the number of samples is 51.
I made 40 sound groups. ATRAC the sound group whose sample number is 511
When encoding with 5 in the previous sound group
The twelfth sample is used as dummy data. One dummy data is added, and the sound group having 512 samples is encoded by ATRAC.

Next, reproduction will be described. At the time of reproduction, that is, at the time of decoding, of the 512 samples output from the ATRAC decoder, the latter 511 samples may be used as actual data. The first sample was discarded because it was treated as dummy data during encoding. The boundaries of the sound group, as described below,
Since there is an overlap and the process of multiplying by the window function is performed, the continuity of the waveform is maintained. Therefore, in the processing of one sample serving as dummy data, it can be considered that the number of overlap samples increases by one sample as compared with the related art. However, it is necessary to record which sound group outputs 511 samples or 512 samples in the header information at the time of encoding.

As shown in FIG. 6, the recording medium has GOP header information and 1 frame consisting of 15 frames as described above.
Video data of GOP (about 0.5 seconds) and audio data of 47 sound groups corresponding to about 0.5 seconds (1 GOP) are handled as minimum units. At this time,
One sound group is composed of 511 or 512 samples, and there are 24024 samples in one GOP. The video data composed of 15 frames is M
The compression processing is performed by PEG2, and the audio data composed of 47 sound groups is compression-processed by ATRAC. In addition, information on the position of the sound group in which the number of valid audio data samples is set to 511 by exception processing is stored in the header information. The header information and the recording data (bit stream) composed of the compressed video data and audio data are modulated by (1,7) RLL,
It is recorded on a recording medium.

Video data and audio data are recorded on this recording medium in units of 1 GOP. First, following the header information, compressed video data for one GOP, and subsequently, audio data for two channels are recorded in a period of one GOP. In the header information, which sound group is 51
The number of one or 512 samples is recorded. The recorded audio data is (47 (SG) × 2 (c
h) =) Consists of 94 sound groups. Furthermore, GO
The phases of the video data and the audio data always match in P units.

When the designation of the reproduction position is received from the user in GOP units, the bit stream from the designated position is demodulated, and the demodulated bit stream is input to the video data and audio data decoder. The decoder for video data is the MPEG decoder 44, and the decoder for audio data is the ATRAC decoder 43. A
The output of the TRAC decoder 43 is temporarily input to the memory 45. After being decoded, the audio data is stored in the memory 45, and is output to the subsequent stage after performing timing adjustment with the video data. ATRAC decoder is 4
3 and the output of the MPEG decoder 44 are completed in GOP units, so that once they are synchronized, there is no deviation.

The audio data received from the ATRAC decoder 43 may have 512 samples or 511 samples per sound group. As described above, the sound group to which the exception processing has been performed is used. Is written in the header information for each GOP, and by reading this, audio data in which the number of samples in one GOP is exactly 24024 is obtained.

As described above, in this example, since the video data and the audio data are completed in one GOP, it is possible to perform insertion, dubbing, and the like in GOP units without any problem.

FIG. 4 is a detailed block diagram of the ATRAC encoder 28 shown in FIG. PCM data in which each sample from the memory 26 consists of 16 bits (linear PCM)
Is supplied to the ATRAC encoder 28 via the input terminal 61. This PCM data is supplied to the ring buffer 64. The sample number information of the sound group is supplied from the input terminal 62 to the ATRAC encoder 28, and is supplied from the output terminal 63 to the subsequent multiplexer 30. The information on the number of samples of this sound group is supplied to the address control circuit 65.

The address control circuit 65 determines whether the sound group is composed of 511 samples from the supplied sample number information of the sound group.
It is determined whether the sample consists of 512 samples. If the address control circuit 65 determines that this sound group is a sound group consisting of 511 samples, the start address read from the ring buffer 64 is shifted by one. The ring buffer 64 has a capacity of three sound groups, as shown in FIG. 7A, and stores input PCM data.

The multiplier 66 multiplies the PCM data read from the ring buffer 64 by the window function, as shown in FIG. 7A. In FIG. 7A, a solid line portion represents processing for one sound group this time, and a broken line part represents processing for the next one sound group. The result of the multiplication is supplied to the band division filter 2, and is divided into a high band component (H), a middle band component (M), and a low band component (L). The high frequency component (H) is MDCT (Modified Discrete Cosine T)
ransform) 4, and the other components are supplied to the band division filter 3. In the band division filter 3, the supplied component is divided into a middle band component (M) and a low band component (L). The divided middle frequency component (M) is supplied to MDCT5, and the low frequency component (L) is supplied to MDCT6.

In the MDCTs 4, 5, and 6, the supplied high-frequency component, middle-frequency component, and low-frequency component are subjected to a modified DCT operation and decomposed into frequency components. MDCT4
The outputs of 5 and 6 are output by the bit allocation calculation circuit 8 and the normalization /
It is supplied to the quantization circuit 7. The bit allocation calculation circuit 8 calculates the number of quantization bits to be allocated by calculation from the supplied components. The information indicating the number of quantization bits is supplied to the normalization / quantization circuit 7, output from the output terminal 10, and supplied to the subsequent multiplexer 30.

The normalization / quantization circuit 7 determines the number of quantization bits from the information on the number of quantization bits, and uses the number of quantization bits to determine the high-frequency components and the mid-frequency components supplied from the MDCTs 4, 5, and 6. The quantization is performed on the component and the low-frequency component (AC component). Further, normalization is performed on each quantized AC component. Further, from the normalization / quantization circuit 7, the processed AC component is output as spectrum information together with the DC component.
Is supplied to the multiplexer 30 at the subsequent stage.

Here, the number of audio data samples is 511.
A method of encoding the sound groups by ATRAC will be specifically described. ATRA used in this example
The encoding of C is one sound group or 512
This is executed in units of the number of samples of the audio data.
However, actually, in addition to the 512 samples, 256 samples before and after the sample are added, and a total of 1 sample is obtained.
ATRAC encoding is performed by taking data of 024 samples and multiplying by a window function. This window function is a function that fills both ends of the total number of 1024 samples with 0, as shown in FIG.

Of course, when handling 511 samples, 1
024 samples are used as input to the ATRAC encoder. However, the number of valid samples in this case is reduced by one sample to 511 samples from the 258th to the 768th. That is, as shown in FIG.
In the case of encoding the second sound group, the ATRAC encoder receives a total of 1,024 samples including the number of samples of the central sound group, 512, and the 256 samples before and after the sound group.

However, when encoding the next (n + 1) th sound group, since the number of samples of the sound group is 511, the number of samples including the 512th of the nth sound group is set to 512, and 5
A total of 1,024 of the 256 samples before and after the 12 sound groups is input to the ATRAC encoder. In other words, when the number of samples in the sound group is 511, only one sample is overlapped and processed. This overlapped portion is the original processing in the encoding processing of the n-th sound group. In the process of encoding the (n + 1) th sound group, it is processed as dummy data.

FIG. 5 is a detailed block diagram of the ATRAC decoder 43 shown in FIG. The information on the number of quantization bits and the information on the normalization coefficient are supplied from the input terminal 12 to the ATRAC decoder 43, and similarly, the spectrum information is supplied from the input terminal 11 to the ATRAC decoder 43. Also, the number of samples of the sound group is input from the input terminal 71 as AT.
The signal is supplied to the RAC decoder 43 and the output terminal 73
Is supplied to the CPU 32 via the.

The information on the number of quantization bits, the information on the normalization coefficient and the spectrum information supplied to the ATRAC decoder 43 are supplied to the spectrum restoration circuit 13. The spectrum restoration circuit 13 generates an AC component from the supplied spectrum information, information on the normalization coefficient, and information on the number of quantization bits. The high frequency component (H) of the generated AC component is represented by I
It is supplied to an MDCT (inverse MDCT) 14. Similarly AC
The mid-range component (M) of the component is supplied to the IMDCT 15. Further, the AC component and the DC component of the low frequency component (L) are supplied to the IMDCT 16.

In the IMDCTs 14, 15 and 16, inverse MDCT processing is performed on the supplied components. IM
The signal of the high frequency component obtained by the inverse MDCT processing in the DCT 14 is supplied to the band synthesis filter 17. Also, I
The signals of the middle band component and the low band component obtained by the inverse MDCT processing by the MDCTs 15 and 16
Supplied to The band combining filter 18 combines the signal of the middle band component and the signal of the low band component, and the combined signal is
This is supplied to the band synthesis filter 17. The band synthesis filter 17 synthesizes the high frequency component and the synthesized signal, and the synthesized signal is supplied to the multiplier 74.

The multiplier 74 multiplies the window function by the synthesized signal, and the result of the multiplication is recorded in the ring buffer 75. This window function is 1024 as shown in FIG.
This function fills both ends of the total number of samples with zero. As shown in FIG. 7A, when writing the multiplication result obtained by the current window function and the overlap of the multiplication result obtained by the next window function in the ring buffer 75, the previously written data is read. , The read data and the current data are added. The window function thus synthesized is recorded in the ring buffer 75. The ring buffer 75 has a capacity of three sound groups. In the address control circuit 76, the input terminal 7
It is determined from the sample number information of the sound group supplied from 1 whether the sound group recorded in the ring buffer 75 is composed of 511 samples or 512 samples. When the address control circuit 76 determines that this sound group is a sound group composed of 511 samples, the start address of the ring buffer 75 is shifted by one. P consisting of 16 bits from the ring buffer 75
When the CM data is read, one sample of the dummy data is removed, and the memory 4 in the subsequent stage is
5.

According to the video signal and audio signal recording / reproducing apparatus as shown in FIGS. 3 to 7, the video data compressed by MPEG and the audio data compressed by ATRAC are synchronously recorded on a recording medium. be able to.

According to this example, recording is performed in GOP units,
Playback can be performed. At this time, synchronization from video and audio (lip sync) is maintained regardless of which GOP is started.

[0054]

In such a conventional video signal and audio signal recording / reproducing apparatus, one sound group is controlled by controlling the ring buffers 64 and 75 in the ATRAC encoder 28 and the ATRAC decoder 43 from the external CPU 32. The number of samples per hit is appropriately set to 511 or 512 so as to maintain a synchronous relationship.

However, this is because the sound groups of the even number samples and the odd number samples are arranged side by side, and there has been a disadvantage in that, when multiplied by the window function, the amplitude fluctuates in principle. This amplitude fluctuation was about 0.2 dB as a result of the simulation.

In the above-mentioned conventional example, since it is necessary to control the pointers of the ring buffers 64 and 75, the ATRAC encoder and the AT which are currently used for the MD are used.
RAC decoder cannot be used as is (new LSI
Need to make. ) There was an inconvenience.

In view of the above, it is an object of the present invention to avoid the above-mentioned fluctuation of the amplitude and to make it possible to use the ATRAC encoder and ATRAC decoder currently used.

[0058]

The video signal and audio signal recording apparatus of the present invention encodes video data by MPEG and generates compressed video data. A / D conversion means for digitizing the audio signal at this sampling frequency, and audio data from the A / D conversion means being encoded by ATRAC. Audio encoding means for generating compressed audio data of a fixed number of sound groups for the IGOP of the compressed video data, and a dummy for making the number of audio data samples supplied to the audio encoding means constant per sound group Dummy data adding means for adding data, and the compressed video data and the compressed audio In those having a recording means for recording.

According to the present invention, after a video signal is compressed by MPEG and an audio signal is compressed by ATRAC,
Synchronization can be performed in units of IGOP, and can be recorded on a recording medium. In this case, in the present invention, the ATR
The number of samples of audio data supplied to the audio encoding means for encoding by AC is set to a fixed number, for example,
Since two dummy data are added, the number of samples in each sound group is fixed at, for example, 512, so that there is no fluctuation in amplitude when performing synthesis by multiplying by a window function.

According to the present invention, since the number of samples in each sound group is fixed at, for example, 512, it is not necessary to control the pointer of the ring buffer, and the ATRAC encoder and the AT which are used in the conventional MD are used.
The RAC decoder can be used as it is.

[0061]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a video signal and audio signal recording apparatus according to an embodiment of the present invention; The overall block diagram of this embodiment is the same as that of FIG. 3 described above. In this embodiment, the ATRAC encoder 28 of FIG. 3 is configured as shown in FIG. 1, and the ATRAC decoder 43 is configured as shown in FIG.

Since the description of the present embodiment has already been given with reference to the entire block diagram of FIG. 3, it is omitted here.

Here, an example of synchronously recording video data compressed by MPEG2 and audio data compressed by ATRAC according to the present embodiment will be specifically described.

The sampling frequency of the audio signal is GOP
A frequency that is an integral multiple of the frequency (2000/1001 Hz) is set. Specifically, the least common multiple of the GOP frequency and the sampling frequency is small, and 48.0 kHz, which is generally used, is set as the sampling frequency of the audio signal. The number of samples of audio data in one GOP obtained from this sampling frequency is 24024. A
The TRAC encoder defines eight modes, and in this example, a coding mode: AT
RAC (1), Level, Channel No. =
2 will be described.

In this coding mode, the number of 512 samples of one channel of audio data is encoded into 212 bytes of compressed data. The processing unit of ATRAC with the number of 512 samples is called a sound group (SG). Hereinafter, only the L channel will be described for ease of description. 24024
When the number of samples of the audio data is divided into 512 sample units, the number of samples becomes 46 sound groups + 472 audio data. Therefore, if 47 sound groups are prepared, a fraction is generated, but audio data for one GOP can be stored.

When the number of samples of audio data for one GOP sampled at the sampling frequency of 48.0 kHz is compared with 24024 samples of audio data for one GOP, the number of samples of audio data for one GOP is 47 ( Since SG) × 512 (samples) = 24064 (samples), a fraction of 40 samplings is generated for the number of samples of one GOP (24024).

Since the number of samples of the sound group in the currently used ATRAC encoder is 512, in this example, the currently used ATRAC encoder is used as the ATRAC encoder 28, and the ATRAC encoder 28 has 512 per sound group. The audio data of a fixed number of samples is supplied.

In this case, the number of samples of the audio data of one GOP is 24024 as described above, so that 40 data are insufficient, but in this example, the ATRAC encoder 2
As shown in FIG. 1, a dummy data adding means is provided at the preceding stage of FIG. 8 so that these 40 dummy data are added to the audio data at an appropriate timing.

In this case, the sound group consisting of 512 sound data samples can be arranged evenly among the 47 sound groups as much as possible. For example, the 0, 7, 14, 27, 34, 41st A sound group consisting of 512 samples of audio data is arranged in this case, and dummy data is added one by one to the other 40 sound groups. The number of samples is 51
Two fixed unit sound groups and 1G
Audio data of 47 sound groups is ATR for OP
It is supplied to the AC encoder 28 and encoded.

Information in which dummy data is added to the audio data, that is, information as to what part of which sound group the dummy data is added to, is recorded in a separate area such as a header.

Next, reproduction will be described. ATR of this example
In the AC decoder 43, the ATRAC encoder 2
Since the number of samples of audio data is fixed to 512 for one sound group in step 8, the ATRAC decoder side also needs to perform the same processing (in this case, the existing A
The TRAC decoder can be used as it is. ).

Therefore, the number of samples of 47 (SG) × 512 (samples) = 24064 (samples) per GOP is obtained on the output side of the ATRAC decoder 43. If the dummy data is removed from the 24064 samples according to the dummy data additional information, audio data of 24024 samples of valid data per GOP can be obtained.

To remove the dummy data in this case, FIG.
When reading audio data from the memory 45 of the CPU 3
2 may be removed by a control signal.

As shown in FIG. 6, the recording medium has GOP header information and 1 frame consisting of 15 frames as described above.
Video data of GOP (about 0.5 seconds) and audio data of 47 sound groups corresponding to about 0.5 seconds (1 GOP) are handled as minimum units. At this time,
One sound group consists of 512 samples,
There are 24064 samples in one GOP. Then, video data composed of 15 frames is subjected to compression processing by MPEG2, and audio data composed of 47 sound groups is subjected to compression processing by ATRAC. Also, information on the position of the sound group to which the dummy data has been added is stored in the header information. The header information and the recording data (bit stream) composed of the compressed video data and audio data are modulated by (1,7) RLL,
It is recorded on a recording medium.

Video data and audio data are recorded on this recording medium in units of 1 GOP. First, following the header information, compressed video data for one GOP, and subsequently, audio data for two channels are recorded in a period of one GOP. In the header information, which position of which sound group the dummy data is added to is recorded.
The recorded audio data is (47 (SG) × 2 (c
h) =) Consists of 94 sound groups. Furthermore, GO
The phases of the video data and the audio data always match in P units.

When a designation of a reproduction position is received from the user in GOP units, the bit stream from the designated position is demodulated, and the demodulated bit stream is input to a video data and audio data decoder. The decoder for video data is the MPEG decoder 44, and the decoder for audio data is the ATRAC decoder 43. A
The output of the TRAC decoder 43 is temporarily input to the memory 45. After being decoded, the audio data is stored in the memory 45. The audio data is output to the subsequent stage after synchronizing the timing with the video data and removing the dummy data. ATRAC decoder 43 and MPEG decoder 4
Since the output of 4 is completed in GOP units, there is no shift once synchronization is achieved.

The audio data received from the ATRAC decoder 43 has 512 samples per sound group. However, as described above, when reading from the memory 45, 40 dummy data are removed and 1 GOP
The audio data in which the number of samples in is exactly 24024 is obtained.

As described above, in this example, since the video data and the audio data are completed in one GOP, it is possible to perform insertion, dubbing, and the like in GOP units without any problem.

FIG. 1 shows a detailed block diagram of the ATRAC encoder 28 in FIG. PCM data in which each sample from the memory 26 consists of 16 bits (linear PCM)
Is supplied to the input terminal 61.

In this embodiment, the 16-bit CPM data supplied to the input terminal 61 is supplied to one fixed contact 80a of the changeover switch 80, and the other fixed contact 80a is supplied via the delay element 81 for one sample. 80b, and the movable contact 80c of the changeover switch 80 is switched by the dummy data additional information from the CPU 32 supplied to the dummy data additional information input terminal 82. The movable contact 80c has 512 samples per sound group. Try to get a number of PCM data.

The PCM data obtained at the movable contact 80c is supplied to the ring buffer 83. Further, the sound group dummy data additional information from the dummy data additional information input terminal 82 is supplied from the output terminal 85 to the subsequent multiplexer 30.

The PCM data of 512 samples per sound group from the ring buffer 83 is supplied to the band division filter 2 via the multiplier 84.

The multiplier 84 multiplies the PCM data read from the ring buffer 83 by the window function. The result of the multiplication is supplied to the band division filter 2, and is divided into a high band component (H), a middle band component (M), and a low band component (L). The high-frequency component (H) is a MDCT (Modified Discret
e Cosine Transform) 4 and the other components
The signal is supplied to the band division filter 3. Band division filter 3
In, the supplied component is divided into a middle band component (M) and a low band component (L). The divided mid-range component (M) is M
The DCT 5 supplies the low frequency component (L) to the MDCT 6.

In the MDCTs 4, 5, and 6, the supplied high-frequency component, middle-frequency component, and low-frequency component are subjected to a modified DCT operation and decomposed into frequency components. MDCT4
The outputs of 5 and 6 are output by the bit allocation calculation circuit 8 and the normalization /
It is supplied to the quantization circuit 7. The bit allocation calculation circuit 8 calculates the number of quantization bits to be allocated by calculation from the supplied components. The information indicating the number of quantization bits is supplied to the normalization / quantization circuit 7, output from the output terminal 10, and supplied to the subsequent multiplexer 30.

The normalization / quantization circuit 7 determines the number of quantization bits from the information on the number of quantization bits, and uses the number of quantization bits to calculate the number of high-frequency components supplied from the MDCTs 4, 5, and 6, Quantization is performed on the band component and the low band component (AC component). Further, normalization is performed on each quantized AC component. Further, from the normalization / quantization circuit 7, the processed AC component is output as spectrum information together with the DC component.
Is supplied to the multiplexer 30 at the subsequent stage.

The ATRAC encoding used in this example is executed in units of one sound group, that is, 512 audio data samples. However, actually, in addition to the 512 samples, 256 samples before and after the sample are added, and data of a total of 1024 samples is obtained.
AC encoding is performed. This window function is 102
It is a function that fills both ends of all four samples with zeros.

FIG. 2 is a detailed block diagram of the ATRAC decoder 43 shown in FIG. The information on the number of quantization bits and the information on the normalization coefficient are supplied from the input terminal 12 to the ATRAC decoder 43, and similarly, the spectrum information is supplied from the input terminal 11 to the ATRAC decoder 43. Further, the dummy data additional information of the sound group is supplied from the input terminal 86 to the CPU 32 via the output terminal 87.

The information on the number of quantization bits, the information on the normalization coefficient and the spectrum information supplied to the ATRAC decoder 43 are supplied to the spectrum restoration circuit 13. The spectrum restoration circuit 13 generates an AC component from the supplied spectrum information, information on the normalization coefficient, and information on the number of quantization bits. The high frequency component (H) of the generated AC component is represented by I
It is supplied to an MDCT (inverse MDCT) 14. Similarly AC
The mid-range component (M) of the component is supplied to the IMDCT 15. Further, the AC component and the DC component of the low frequency component (L) are supplied to the IMDCT 16.

In the IMDCTs 14, 15, and 16, inverse MDCT processing is performed on the supplied components. IM
The signal of the high frequency component obtained by the inverse MDCT processing in the DCT 14 is supplied to the band synthesis filter 17. Also, I
The signals of the middle band component and the low band component obtained by the inverse MDCT processing in the MDCTs 15 and 16 are combined with the band synthesis filter 18.
Supplied to The band combining filter 18 combines the signal of the middle band component and the signal of the low band component, and the combined signal is
This is supplied to the band synthesis filter 17. The band combining filter 17 combines the high-frequency component and the combined signal, and the combined signal is supplied to the multiplier 88.

The multiplier 88 multiplies the window function by the synthesized signal, and the result of the multiplication is recorded in the ring buffer 89. This window function is a function that fills both ends of the total number of 1024 samples with zero.

The 16-bit PCM data from the ring buffer 89 obtained at the output terminal 90 is supplied to the memory 45, and the dummy data additional information from the output terminal 87 is supplied to the CPU 32. In accordance with the control signal from the CPU 32, The timing of the video data is adjusted from the memory 45 and audio data from which the dummy data is removed is obtained. Otherwise, the same operation as described with reference to FIG. 3 is performed.

According to this embodiment, the video signal is compressed by MPEG and the audio signal is compressed by ATRAC,
Synchronization is achieved in units of P, and recording can be performed on a recording medium.

In this case, in this example, the number of samples of the audio data supplied to the ATRAC encoder 28 to be encoded by ATRAC is set to a fixed 512 per sound group.
Since dummy data is added, the number of samples in each sound group is a fixed 512, and there is an advantage that there is no fluctuation in amplitude when synthesizing by applying a window function.

According to this embodiment, since the number of samples in each sound group is constant at 512, it is not necessary to control the points of the ring buffers 83 and 89, and the ATRAC encoder and ATTR used in the conventional MD are used.
There is an advantage that the AC decoder can be used as it is.

Further, in this example, since the compressed video data and the compressed audio data are completed in GOP units as described above, it is possible to perform video replacement and audio post-recording editing using one GOP as a minimum unit. .

Further, since the audio data is stored as 47 sound groups in one GOP, the editing of the audio data is not affected at all without affecting other places.
Editing can be performed in sound group units (about 10 msec). However, in this case, whether the number of valid samples is 511 or 512 depends on the number of samples of the audio data recorded based on the sample.

In the above-described example, the delay element 81 is provided as dummy data, and the delay element 81 is held beforehand. However, the dummy data may be interpolated from the preceding and succeeding sample data by an average value. For the purpose of ATRAC encoding in the subsequent stage, it is desirable that the addition of the dummy data does not cause unnecessary high-frequency components. In addition, the present invention is not limited to the above-described example, and may adopt various other configurations without departing from the gist of the present invention.

[0098]

According to the present invention, a video signal is compressed by MDEG, and an audio signal is compressed by ATRAC.
Synchronization is performed in units of GOP, and recording can be performed on a recording medium.

In this case, according to the present invention, the number of samples of audio data to be supplied to the audio encoding means for ATRAC encoding is fixed at, for example, 512 per dummy data group. The number of samples in a group is a fixed number, for example, 512, and there is an advantage that amplitude does not fluctuate when combining by multiplying by a window function.

According to the present invention, since the number of samples in each sound group is fixed at, for example, 512, there is no need to control the point of the ring buffer.
There is an advantage that the ATRAC encoder and ATRAC decoder used in the above can be used as they are.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of an ATRAC encoder used in an example of a video signal and audio signal recording device according to the present invention.

FIG. 2 is a block diagram illustrating an example of an ATRAC decoder used in the example of FIG. 3;

FIG. 3 is a block diagram showing an example of a video signal and audio signal recording / reproducing apparatus to which the present invention is applied.

FIG. 4 is a block diagram illustrating an example of a conventional ATRAC encoder.

FIG. 5 is a block diagram illustrating an example of a conventional ATRAC decoder.

FIG. 6 shows video data and ATRAC compressed by MPEG.
FIG. 3 is a diagram for explaining audio data to be compressed by the following.

FIG. 7 is a diagram for explaining a conventional ATRAC encoder;

FIG. 8 is a block diagram illustrating an example of a conventional ATRAC encoder.

FIG. 9 is a block diagram illustrating an example of a conventional ATRAC decoder.

[Explanation of symbols]

1: input terminal, 2, 3: band division filter, 4, 5, 6
... MDCT circuit, 7 ... Normalization / quantization circuit, 8 ... Bit allocation calculation circuit, 9,10 ... Output terminals, 11,12 ...
Input terminal, 13: spectrum restoration circuit, 14, 15, 1
6 ... IMDCT circuit, 17, 18 ... band synthesis filter,
19: output terminal, 21: video camera, 22, 23, 2
4: A / D converter, 25: Clock generator, 26, 45
... Memory, 27 ... Camera process circuit, 28 ... ATRA
C encoder, 29: MPEG encoder, 30: multiplexer, 31: channel coding circuit, 32
... CPU, 33 ... recording medium, 41 ... channel decoding circuit, 42 ... demultiplexer, 43 ... ATRA
C decoder, 44: MPEG decoder, 46: video encoder, 47: digital filter, 49: D / A converter, 53: TV monitor, 61, 62, 71, 82, 86
... input terminals, 63, 72, 73, 85, 87 ... output terminals, 64, 75, 83, 89 ... ring buffers, 66,
74, 84, 88: multiplier, 80: changeover switch, 81
… Delay element

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁶ Identification code FI H04N 7/24 H04N 7/13 Z

Claims (2)

[Claims] 1. A video encoding means for encoding video data by MPEG to generate compressed video data, and setting a sampling frequency of an audio signal so as to include a fixed number of samples in GOP units of the compressed video data. A / D conversion means for digitizing the audio signal at the sampling frequency; and ATRAC encoding of audio data from the A / D conversion means, and a fixed number of sound groups for one GOP of the compressed video data. Audio encoding means for generating compressed audio data; dummy data adding means for adding dummy data for making the number of samples of audio data supplied to the audio encoding means constant per sound group; The compressed audio data and the GO
A recording device for recording a video signal and an audio signal, comprising: recording means for recording in P units. 2. The video signal and audio signal recording apparatus according to claim 1, wherein information on the number of effective samples of the audio data is recorded in a separate area such as a header. Recording device.