JPH08287612A - Variable speed reproducing method for audio data - Google Patents

Abstract

PURPOSE: To reduce a noise on acoustic feeling in the case of conducting a variable speed reproduction of an audio signal without changing an interval from that at the time of a normal reproduction. CONSTITUTION: Audio data are compressed every audio data block having a prescribed time length, and the audio data recorded by a compression system enhancing continuity of a sound waveform is reproduced so that respective data block having the prescribed time length are overlapped. Block groups GR1-GR4 of such number of blocks that plural pieces of the audio data blocks of decodable minimum unit are continued and reproduction repeating frequency becomes one with low sensing sensitivity on the auditory sense of human when reproduction is repeated with plural pieces of blocks as a unit are generated from the recorded audio data hourly continuing. The reproduction at a speed higher than the normal reproducing speed is performed by performing data expansion processing and consecutive processing of an overlap part by block group unit.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a variable speed reproduction method of audio data suitable for application to, for example, a so-called mini disk device.

[0002]

2. Description of the Related Art A so-called mini disk (hereinafter referred to as a small optical disk) is known as a recording medium in which an audio signal is compressed as digital data and recorded. The compression method used for this small optical disc is
ATRAC (Adaptive TRans) in consideration of human auditory characteristics
It is called form acoustic coding. This audio data compression method performs a compression process in units of audio data blocks of a predetermined time length. However, DCT (Discrete Cosine Transform) and a window function are successfully combined so that adjacent blocks are The MDCT (Modified D
iscrete Cosine Transform) is called.

Specifically, when recording on a small optical disk, the digital audio data is divided into unit blocks called sound frames for each 512 samples, and the window processing is performed on the sound frames to create a space between adjacent sound frames. So that they overlap. Then, in this sound frame, DCT is performed to convert the time axis data into the frequency axis data and compress the data.

Then, a block called 1 sound group is generated by 2 sound frames, and 5.5 sound groups are recorded on the disk as 1 sector. When the audio is two-channel stereo, two-channel data corresponds to two sound frames, so that the sound frame and the sound group have the same time length. In other words, if the sampling frequency is 44.1 kHz, 512 samples are 11, 61 msec, and 1
One sound frame includes audio data for one channel of this time length, and a sound group includes audio data for two stereo channels of the same time length.

Therefore, in a small-sized optical disk, the unit that can be encoded / decoded is 11.61 msec worth of audio data, but if the audio data is monaural 1 channel, it is a sound frame, and 2 With channel stereo,
It's a sound group. In this specification, in order to avoid complexity, the following description will be made assuming that the audio data is 2-channel stereo, and the unit that can be encoded / decoded is a sound group.

In reproducing the audio data,
Sound group data is played back from the data picked up from the disc, and inverse MDCT is performed on the sound group data, and the connection processing of the overlap portion with the preceding and succeeding sound groups is performed using the window function to obtain a good sound. Play back.

By the way, conventionally, various techniques for varying the reproduction speed without changing the pitch are provided not only in the field of digital audio. In this case, the voice is generally recorded and stored in the form of time domain data.
Therefore, for this audio data, in order to realize a change in the playback speed without changing the pitch, the time domain data is once converted into the frequency domain data, and then some processing is performed. The process of returning to the time domain data again is required.

As described above, in order to perform the above processing,
Since it is necessary to perform a troublesome process such as conversion between the time domain and the frequency domain, a method of simply performing the above operation has been proposed. This is a process of reproducing a certain region and skipping or superimposing an appropriate region, and it is possible to reproduce a frequency that can be represented only by the certain region.

[0009]

By the way, as described above, the compression method A used in small optical disks is used.
In TRAC, the unit of encoding / decoding is
Since it is in units of sound groups, the playback speed can be changed by performing discrete playback in units of this sound group.

However, in such a case,
Bass noise (boost) occurs every about 86 Hz, which is the period of the sound group. Because, in the connection processing by the window function described above, each window processing unit has noise after inverse MDCT,
This is because the window function is designed so as to cancel the noise when the processing of connecting the overlapping portions is performed between the sound groups that are adjacent to each other.

In other words, it is designed so that the original waveform can be obtained by connecting consecutive sound groups.
In the case of discontinuous sound groups as described above, noise is generated for each discontinuity, that is, for each sound group. Since the noise level does not change so much, it results in a boost in the low range.

The above problem is not limited to the case of reproducing a small-sized optical disk, but is that the audio data is compressed in block units divided into predetermined time lengths, and the blocks before and after that are overlapped. This is a common problem in the case of variable speed reproduction of audio data recorded / saved after processing.

In view of the above points, the present invention reduces the absolute amount of noise as described above and performs S / H on human hearing when performing variable speed reproduction without changing the pitch.
It is an object of the present invention to provide a reproducing method capable of improving N.

[0014]

In order to solve the above problems, in the variable speed reproduction method of audio data according to the invention of claim 1, the audio data blocks are compressed for each predetermined time length. When reproducing audio data recorded by a compression method that increases the continuity of the audio waveform by overlapping the data blocks for each predetermined time length, the smallest unit that can be decoded from the recorded audio data that is temporally continuous. Of audio data blocks in a row, and when repeating with a plurality of blocks as a unit, a block group having a number of blocks whose repetition frequency is low for human hearing , The data in the middle is thinned out and repeatedly extracted in sequence, and whether the extracted continuous blocks are The composed blocks, as are continuous sequentially in time, it performs a data decompression process and the connecting process of the overlapped portion, characterized in that to perform the high-speed reproduction than the normal playback speed.

In the audio data variable speed reproducing method according to the present invention, the audio data blocks are compressed for each predetermined time length, and the data blocks for each predetermined time length overlap. When reproducing audio data recorded by a compression method that enhances the continuity of the audio waveform in this way, the minimum decodable audio data block is a plurality of consecutive audio data blocks. When a block group having a number of blocks whose repetition frequency is low for human hearing, the data decompression process is performed as a continuous reproduction of the same block. By connecting the overlapping part, the Characterized in that to perform live.

Further, in the audio data variable speed reproduction method according to the present invention, the audio data is compressed for each block of a predetermined time length, and the data block for each predetermined time length is over. When playing back audio data that has been compressed and recorded by increasing the continuity of the audio waveform by wrapping it, the recorded audio data that is temporally continuous consists of a series of minimum decodable audio data blocks. Then, when the plurality of blocks are repeated as a unit, the repetition frequency is divided for each block group of the number of blocks such that the human auditory sense has low sensitivity, and the block group is divided into It overlaps with the blocks before and after, and the amount of overlap depends on the playback speed. Performs fast playback from the normal playback in the variable, by providing the temporal space between said blocks, characterized in that to carry out the low-speed reproduction than normal playback.

[0017]

According to the present invention having the above-described structure, the process for connecting is performed for the block group consisting of a plurality of continuous blocks, and when the process for connecting is performed in discrete discrete block units. In comparison, the number of appearances of noise is reduced. In addition, the frequency of the noise is driven to the low frequency side where the perception sensitivity is low in human hearing.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention is applied to the recording / reproducing apparatus for the above-mentioned small optical disk as an example.
This will be described with reference to the drawings. First, before describing the present invention, a recording / reproducing apparatus to which the present invention is applied will be described.

FIG. 3 shows the structure of a recording / reproducing apparatus for a small optical disk to which the present invention is applied. This figure 3
In the figure, 1 is a small optical disk. The small optical disc 1 is a disc 1 having a diameter of 64 mm in the cartridge 1A.
It is configured to accommodate B. The small optical disk 1 includes a read-only optical disk, a recordable magneto-optical disk,
There are three types of hybrid discs in which a read-only area and a recordable area are mixed.

Further, a pre-groove for light spot control (for tracking control) is formed (pre-pit) on the disc 1B in advance. Particularly, in this example, this pre-groove is used for tracking. Absolute address data is recorded so as to be superimposed on the wobbling signal.

The disc 1B of the compact optical disc 1 is rotated by the spindle motor 2. Spindle motor 2
Is controlled by the servo control circuit 5 so that the disk 1B rotates at a constant linear velocity.
The small optical disk 1 is provided with a shutter, and when the small optical disk 1 is placed on the disk mounting tray and loaded in the apparatus, the shutter is opened. In the case of a recordable optical disc, the magnetic head 3 for recording is disposed above the shutter opening of the disc 1B, and the optical pickup 4 is disposed below the shutter opening of the disc 1B. Are arranged.

The movement of the optical pickup 4 is controlled by the feed motor 6 in the radial direction of the disc 1B. Further, the servo control circuit 5 controls focus and tracking of the optical pickup 4.

The system controller 20 is constructed by mounting a microcomputer and manages the entire operation. A key input signal is applied to the system controller 20 from the key group 10. This key group 10
Includes a power key, an eject key, a play key, a pause key, a stop key, a recording key, a fast-forward play key, a fast-reverse play key, and the like.

A display 30 is connected to the system controller 20. The display 30 displays time information such as the total playing time of the mounted small optical disk, the elapsed time of the song being played, the remaining playing time of the song being played, the remaining playing time of the entire song, and the information of the song being played. Track number etc. are displayed. In addition, the disc name and the track name are displayed on the disc on which the disc name and the track name are recorded. Furthermore, if the recording date and time of the song or disc is recorded, the recording date and time is displayed.

The structure of the recording / reproducing signal system of the embodiment of FIG.
It is devised so that the structure can be simplified as much as possible by using an IC. It should be noted that, during recording and during reproduction, the mode of each unit is switched by a mode switching signal from the system controller 20.

[Description of Recording System] An audio signal (in the figure, for simplification, one channel, but actually two channels stereo. The same applies hereinafter) is input through an input terminal 31. The audio signal from the input terminal 31 is digitized by the A / D converter 32 at a sampling frequency of 44.1 kHz and a quantization bit number of 16 bits.

This digital audio signal is supplied to the audio compression encoding / decoding circuit 33. In the audio compression encoding / decoding circuit 33, the above-mentioned AT
Data compression processing by RAC is performed, and the audio signal is data compressed to about 1/5. That is, the audio signal is data-compressed by using the MDCT which is the DCT process considering the overlap between the adjacent coding units as described above.

In this case, in the circuit 33, the frequency band of the original analog audio signal is low (0 to 5.5).
125 kHz), mid range (5.5125 kHz-11.0)
25 kHz), high range (11.025 kHz to 22.05)
divided into 3 bands of (kHz) and MDCT for each band
Is processed.

Data of 512 samples, which is a compression processing unit, is assigned 256 samples in the high band and 128 samples in the middle band and the low band, respectively. Then, two modes, a short mode and a long mode, are prepared so that the size of the MDCT processing unit can be selected in two ways for each band unit depending on the amount of change in the audio signal.

That is, the short mode is set in the portion where the sound changes drastically with time, and the long mode is set in the portion where the stable waveform is repeated. In the short mode, the number of samples in the MDCT processing unit is 32 samples in common in the three bands, and MDCT is performed on blocks of each 32 samples. In the long mode, the total number of samples in each band, that is, 25 in the high range.
The MDCT processing unit is 6 samples, and 128 samples each in the middle band and the low band.

In the MDCT, for blocks of the processing unit sample number, windows are set so that adjacent blocks overlap each other, and each block is subjected to window processing, and the processing result is subjected to DCT processing.
As shown in FIG. 4, in each frequency band, a window for short mode and a window for long mode are set and processing is performed. In FIG. 4, thick solid lines indicate window waveforms in the long mode, and dotted lines indicate window waveforms in the short mode.

As can be seen from FIG. 4, the time length of the processing unit in the long mode is 11.6 msec which is equal to the sound frame in any frequency band. On the other hand, the time length of the processing unit in the short mode is 1.45 msec in the high range and 2.9 msec in the middle range and the low range.
Becomes

As described above, by providing the two modes, efficient encoding can be performed in a portion where the sound changes drastically with time or a portion where the stable waveform is repeated.

The data of each sound frame includes
Data for each of the three frequency bands is recorded with identification data indicating in which mode the MDCT process is performed.

The audio signal compressed by the audio compression encoding / decoding circuit 33 is stored in the memory controller 34.
Is stored once in the buffer memory 35 controlled by the memory controller 34 via. In the case of this example, as the buffer memory 35, a DRAM having a data capacity of 4 Mbits is used.

The memory controller 34 sequentially reads the compressed data from the buffer memory 35 at a transfer speed of about 5 times the write speed, unless a track jump in which the recording position on the disk 1B jumps due to vibration or the like during recording. , And transfers the read data to the sector-structure data encoding / decoding circuit 36.

When it is detected that a track jump has occurred during recording, the memory controller 34 stops the data transfer to the data encoding / decoding circuit 36 and the compressed data from the audio compression encoding / decoding circuit 33. Are stored in the buffer memory 35. Then, when the recording position is corrected, control is performed to restart the data transfer from the buffer memory 35 to the data encoding / decoding circuit 36.

As will be understood from the above, as the data capacity of the buffer memory 35 in this case, compressed data corresponding to the time from the occurrence of the track jump to the correct correction of the recording position can be stored. Minimum capacity required. In this example, the buffer memory 35 has a capacity of 4 Mbits as described above, and this capacity is selected with a margin so as to sufficiently satisfy the above conditions.

Further, in this case, the memory controller 34
In this recording, during normal operation, memory control is performed so that the data stored in the buffer memory 35 is reduced as much as possible. That is, when the data amount of the buffer memory 35 becomes equal to or more than a predetermined amount, a predetermined amount of data, for example, 32 sectors (1 sector corresponds to 1CD-
Only the data of the ROM sector (which is about 2 Kbytes) is read from the buffer memory 35, and the memory control is performed so as to always secure the write space of a predetermined data amount or more.

The data encoding / decoding circuit 36 is
Compressed data transferred from the buffer memory 35 to C
Encode into sector structure data of D-ROM. 1
A sector contains 5.5 sound groups. In this case, as described above, at the beginning of the data of each sound frame, the MDC in units of either short mode or long mode is set for each of the three frequency bands.
Identification information indicating whether T has been applied is included.

It should be noted that recording / reproduction of audio data is performed in three steps.
2 sectors of audio data (about 2 seconds of the original analog audio signal, but about 0.4
This is done in units of seconds. This 32
The audio data for a sector is hereinafter referred to as a cluster.

The output data of the data encode / decode circuit 36 is supplied to the EFM and CIRC encode / decode circuit 37. In this circuit 37, data is subjected to error detection / correction coding processing and modulation processing suitable for recording, in this example, EFM (8-14 modulation) processing. The code for error detection and correction is C of the CD in this example.
IRC (Cross Interleave Reed-Solomon Code) with interleave changed is used.
The recording data is intermittent data, and a total of 4 sectors (hereinafter referred to as linking sectors) for cluster connection are added before and after 32 sectors of audio data as one cluster to form a unit of 36 sectors. Recorded data.

The recording data thus formed is
It is supplied to the recording magnetic head 3 via the head drive circuit 38. As a result, the magnetic field modulated by the recording data is applied to the disc 1B (magneto-optical disc). Further, the laser beam from the optical pickup 4 is applied to the disc 1B. At the time of this recording, the recording track is irradiated with laser light having a constant power larger than that at the time of reproduction.
By this light irradiation and the modulation magnetic field by the magnetic head 3,
Data is recorded on the disk 1B by thermomagnetic recording. Thus, about 2 seconds (1 cluster) of data of the original audio signal is recorded on the disc 1B in about 0.4 seconds.

The magnetic head 3 and the optical pickup 4 are constructed so as to be movable in the radial direction of the disk 1 in synchronization with each other.

At the time of this recording, the output of the optical pickup 4 is supplied to the address decoder 40 via the RF amplifier 39, and the absolute address wobble-recorded in the pre-groove provided along the track of the disc 1B. The data is extracted and decoded. Then, the detected absolute address data is EFM and CIRC.
It is supplied to the encoding / decoding circuit 37, inserted into the recording data, and recorded on the disc. Also, the absolute address data is supplied to the system control circuit 20 and used for recognition of the recording position and position control.

Further, a signal from the RF amplifier 39 is supplied to the servo control circuit 5, a control signal for constant linear velocity servo of the spindle motor 2 is formed from the signal from the pre-groove of the disk 1B, and the spindle motor 2 is Speed controlled.

[Description of Playback System] Next, the playback will be described. That is, during this reproduction, the spindle motor 2 is controlled by the servo control circuit 5 in the same manner as during recording.
The signal from the pre-groove controls the rotation speed of the disk 1 at the same constant linear velocity as during recording.

At the time of reproduction, the optical pickup 4 detects the reflected light of the laser light applied to the target track,
For example, a focus error is detected by an astigmatism method, a tracking error is detected by, for example, a push-pull method, and a difference in polarization angle (Kerr rotation angle) of reflected light from a target track is detected to reproduce a reproduction RF signal. Is output.

The output of the optical pickup 4 is the RF amplifier 3
9. The RF amplifier 39 is the optical pickup 4
The focus error signal and the tracking error signal are extracted from the output of the above and supplied to the servo control circuit 5, and the reproduced signal is binarized and supplied to the EFM and CIRC encode / decode circuit 37.

The servo control circuit 5 controls the focus of the optical system of the optical pickup 4 so that the focus error signal becomes zero, and the servo control circuit 5 controls the optical system of the optical pickup 4 so that the tracking error signal becomes zero. Performs tracking control.

The output of the RF amplifier 39 is also supplied to the address decoder 40, which extracts and decodes absolute address data from the pregroove. Then, the absolute address data from the decoder 40 is supplied to the system control circuit 20 via the circuit 37 and used for the servo control circuit 5 to control the reproduction position of the optical pickup 4 in the disc radial direction. In addition, the system control circuit 20
The sector unit address information extracted from the reproduction data can also be used to manage the position on the recording track being scanned by the optical pickup 4.

During this reproduction, as will be described later, the disc 1
The compressed data read from B is written in the buffer memory 35, read and expanded, but due to the difference in the transmission rate of both data, the data read by the optical pickup 4 from the disk 1B is performed by the buffer memory 3, for example.
It is performed intermittently so that the data stored in 5 does not fall below a predetermined amount.

In the EFM and CIRC encoding / decoding circuit 37, the signal supplied through the RF amplifier 39 is EFM demodulated and subjected to error correction processing. The output of the EFM / CIRC encode / decode circuit 37 is supplied to a sector-structure data encode / decode circuit 36 to decompress the sector structure of the CD-ROM and decode the data into original data in a compressed state.

The output of the data encode / decode circuit 36 is temporarily stored in the buffer memory 35 via the memory controller 34. And the memory controller 3
In No. 4, if a track jump in which the reproduction position is skipped due to vibration or the like does not occur during reproduction, the compressed data from the circuit 36 is sequentially read and read at a transfer speed of about ⅕ times the write speed. The data is transferred to the audio compression encoding / decoding circuit 33. In this case, the memory controller 34 controls writing / reading of the buffer memory 35 so that the amount of data stored in the buffer memory 35 does not fall below a predetermined value.

When it is detected that a track jump has occurred during reproduction, writing of data from the data encode / decode circuit 36 to the buffer memory 35 is stopped and only data transfer to the circuit 33 is performed. Then, when the reproduction position is corrected, control is performed to restart the data writing from the circuit 36 to the buffer memory 35.

Further, as described above, the memory controller 34 performs memory control during normal operation so that the buffer memory 35 stores as much predetermined data as possible over the minimum necessary. For example, when the data amount of the buffer memory 35 becomes equal to or less than a predetermined amount, the optical pickup 4 intermittently takes in the data from the disc 1B, and the data encoding / decoding circuit 36 is performed.
The memory control is performed such that the data is written from the memory and the read space of a predetermined data amount or more is always secured.

The time required to read the data into the buffer memory 35 is about 0.9 seconds, which corresponds to about 3 seconds of audio data. That is, even if the signal of the disk 1B cannot be read while the buffer memory is full of data, about 3
It is possible to continue outputting the reproduction signal for a second. By re-accessing the optical pickup to its original position and reading the signal again during that time, the occurrence of sound skip can be prevented.

Audio compression encoding / decoding circuit 33
Performs inverse MDCT processing to decompress and band-combines data in three frequency bands. Decompression,
The band-combined data is supplied to the D / A converter 41 and returned to an analog signal. This analog signal is output from the output terminal 42.

[Explanation of Embodiment of Variable Speed Reproduction] In this embodiment, variable speed reproduction is performed by the system controller 20.
And is executed as follows. 1 and 2 are views for explaining the first embodiment of the variable speed reproduction. FIG. 1 shows an example in which reproduction is performed at higher speed than normal reproduction, and FIG. 2 is lower than normal reproduction. Examples of the case of performing the reproduction of are shown respectively.

In the case of this example, when the user operates the variable speed reproduction key, for example, the system controller 20
1A or 2A, data DA on the disk which is continuous in time is shown in FIG. 1A and FIG. The blocks are divided into block groups GR1, GR2, GR3, ... Each group SG.

In this case, each block group GRi (i = 1,
The number of sound groups constituting 2, ...) is set such that the repetition frequency of the block group GRi is low in human auditory sense sensitivity, for example, 20 Hz or less. As described above, when the time length of one sound group is stereo, it is 11.61 msec, and its repetition frequency is 86 Hz. Therefore, in this example,
A group of 4 to 10 consecutive sound groups is defined as one block group GRi. In the example of FIGS. 1 and 2, one block group GRi is composed of five consecutive sound groups SG.

In the case of performing high speed reproduction, as shown in FIG.
As shown in B, a part of each block group GRi which is temporally before and after is overlapped, and the overlap amount is changed according to the selected speed. For example, if one block group GRi is composed of five sound groups as in this example, in order to make the block group GRi 20% faster than the normal speed, the block group GRi should be arranged before and after that, as shown in FIG. 1B. The block group GRi-1 and the block group GRi + 1 are overlapped by one sound group SG.

In FIG. 1B and FIG. 2B, the mountain-shaped curve waveform shows a window waveform, and for convenience sake, only the long mode waveform is shown to show a continuous or discontinuous state.

It should be noted that the overlap amount may not be a unit of one sound group but may be a unit of one sound group or less according to the speed change for normal reproduction. That is, in the example of FIG. 1, if the speedup is 20% or less, the overlap is the time length of one sound group or less.

In this case, in each block group GRi,
Since the sound group data is temporally continuous, noise-free reproduction is performed by performing window processing and overlap processing as in normal reproduction. Then, the connecting process for variable speed reproduction is performed in the overlapping portion between the block groups. This is because processing is performed by a window function that is completely different from that during normal playback, or during variable speed playback, sound quality does not matter so much. Alternatively, a simple process for cross-fading such as so-called fade-in / fade-out may be performed.

Next, when performing the low speed reproduction, as shown in FIG.
As shown in, reproduction is performed with a gap between each block group GRi for a time corresponding to the selected speed. For example, to make the speed 20% slower than the normal speed,
As shown in, the block group GRi is left for one sound group time. If the speed is reduced by 20% or less, the time space between the block groups GRi becomes the time length of one sound group SG or less.

In the case of this low speed reproduction, the block group GR
Since there is no overlap between i, window processing need only be performed using the window function during normal reproduction. However, the data of the front and rear block groups GRi may be cross-faded so as to fill the time space.

As described above, the MDCT window becomes discontinuous every block unit, that is, at every length of four or more continuous sound groups, and the number of noises generated is one. Compared to the case of performing variable speed playback for each sound group, When variable speed reproduction is performed in units of one sound group and the sound is discontinuous, noise of about 86 Hz is generated.
The noise is about 20 Hz or less, which is below, and the noise can be driven to a frequency range where human sensitivity is low.

The audio compression encoding / decoding circuit 33 in the case of this example has a functional block configuration as shown in FIG.

That is, in this example, the spectrum restoration processing section 50 restores the spectrum signal from the input data from the memory controller 34. From the spectrum restoration processing unit 50, data of three frequency bands of high band, middle band, and low band can be obtained.

Then, the high-range, middle-range, and low-range spectrum signals are processed by the inverse MDCT processing units 51, 5 for the respective frequency bands.
2, 53, and inverse MDCT processing units 54, 55, 56
And each of them are supplied.

The inverse MDCT processing units 51 to 56 perform inverse processing of MDCT at the time of recording, and restore the spectrum signal of each band from the frequency domain to the time domain data, that is, the time series signal and decompress the data. Perform processing. Further, at the time of this inverse conversion process, under the control of the system controller 20, a window function corresponding to the short mode or the long mode is generated, and the generated window function is multiplied. In other words, processing is performed so that overlapping portions of adjacent processing unit blocks have a smoothly connected waveform when they are added together. Then, the addition processing of the overlapping portion is also performed.

In this case, the inverse MDCT processing units 51, 52,
53 is a normal reproduction reverse MDCT processing unit 54, 55,
Reference numeral 56 is for variable speed reproduction. Both perform the same operation within a sound group or in successive sound groups.

That is, in the short mode, 1 unit is set for 32 samples in one sound group.
As a processing unit block, while returning the frequency domain data to the time domain data, regarding the time domain data,
The blocks are overlapped with each other to perform connection processing. In the long mode, 256 or 12
Using 8 sample units as one processing unit block, the frequency domain data is restored to the time domain data, and the time domain data is joined by overlapping some adjacent sound groups.

The system controller 20 decodes the identification information in the reproduced data of each sound group, determines whether the recording mode is the short mode or the long mode, and responds to the determination result. The respective inverse MDCT processing circuits 51, 5
2, 53 and inverse MDCT processing circuits 54, 55, 56 are controlled. The inverse MDCT processing circuits 51, 52, 53 and the inverse MDCT processing circuits 54, 55, 56 execute inverse MDCT processing according to the respective modes based on this control.

However, the inverse MDCT processing units 54, 55 and 56 for variable speed reproduction are the parts which are discontinuous during variable speed reproduction under the control of the system controller 20.
As described above, the processing of the overlap portion different from that for normal reproduction is performed.

The switch circuits 57, 58 and 59 select the time series signals from the inverse MDCT processing units 51, 52 and 53 during normal reproduction and the inverse MDCT processing unit 5 during variable speed reproduction.
The time series signals from 4, 55 and 56 are selected, and these switch circuits 57, 58 and 59 are switched by a switching signal from the system controller 20.

The band synthesizing filter 61 includes the switch circuit 5
The time-series data in the middle and low frequencies from 8, 59 are synchronously combined. Delay circuit 62
Delays the high-frequency time series data from the switch circuit 57 by the processing time of the band synthesis filter 61. Further, the band synthesizing filter 63 performs a process of synthesizing the time series data from the delay circuit 62 and the time series data obtained by synthesizing the middle band and the low band from the band synthesizing filter 61 in synchronization with each other. The output of the band synthesis filter 63 becomes the reproduction output data of the audio compression encoding / decoding circuit 33.

FIG. 6 is a flow chart of processing at the time of reproduction in the audio compression encoding / decoding circuit 33.

That is, the circuit 33 first restores the spectrum signal (step S1). Next, it is judged whether or not the reproduction state instructed by the user is variable speed reproduction (step S2). If it is not variable speed reproduction, normal inverse MDCT is performed (step S3), band synthesis of three frequencies is performed (step S8), and the synthesized data is output to the D / A converter (step S9). The processing routine ends.

When it is determined in step S2 that the variable speed reproduction is performed, the one block group GRi is set as described above.
It is determined whether or not the processing of the circuit 33 has been completed for the number of sound groups that are defined as (step S4). If it is determined in step S4 that the sound group in the one-block group GRi is still being processed, step S3
After that, the same reverse MDCT processing as in normal reproduction is performed. This is because within one block group GRi, the sound groups are temporally continuous, and the same processing as during normal reproduction may be performed. Next, band synthesis of three frequencies is performed (step S8), and D / A
The data is output to the converter (step S9), and this processing routine ends.

When it is determined in step S4 that the processing of all the sound groups in one block group GRi has been completed, the processing of the next block group is prepared (step S5), and then the variable speed reproduction overrun is performed. It is determined whether or not processing such as lap processing is performed (step S
6) If the variable speed reproduction process is not to be performed, the process proceeds to step S3 and, as described above, the normal reverse MDCT is performed.
, Band synthesis is performed (step S8), output data is output to the D / A converter (step S9),
This processing routine ends.

Further, when performing the overlap processing for variable speed reproduction or the like, the process proceeds from step S6 to step S7, and the inverse MDCT for variable speed reproduction for processing the overlap portion as described above is performed. Then, the process proceeds to step S8, band synthesis is performed, and the output data is D / A.
The data is output to the converter (step S9), and this processing routine ends.

[Second Embodiment of Variable Speed Reproduction] FIGS. 7 and 8 are views for explaining the second embodiment of the variable speed reproduction. FIG. 7 shows reproduction at a higher speed than normal reproduction. FIG. 8 shows an example of the case of performing the reproduction, and an example of the case of performing the reproduction at a speed lower than the normal reproduction. In addition, in FIG. 7, the mountain-shaped curve waveform indicates a window waveform,
For convenience, this shows only the long mode waveform
It shows the state of discontinuity. The window waveform is omitted in FIG. In the second embodiment, one block group GRi is composed of four sound groups SG.

In the case of this example, when high-speed reproduction is instructed by the user, for example, the system controller 20 responds to the designated speed from the data DA on the disk which is continuous in time as shown in FIG. 7A. The optical pickup 4 is controlled so that the extracted data is extracted. In this example, this data extraction is performed in the unit of the block group GRi described above, but in the case of this high-speed reproduction, the block groups GRi are extracted by skipping the sound groups SG by the number corresponding to the speed. Then, each block group GRi is temporally continuous.

For example, when the block group GRi is composed of four sound groups SG, if the speed is double, as shown in FIG. 7B, four sound groups SG of four are grouped into one block group. The GRi are sequentially extracted. Also, if the speed is increased by 25%, then FIG.
Further, as shown in FIG. 7D, the four sound groups SG in the data DA, which are skipped one by one, are sequentially extracted as one block group GRi and arranged so as to be temporally continuous.

Reproduction at a speed slower than the speed during normal reproduction is performed as shown in FIG. That is, in the case of this example, as shown in FIG. 8A, temporally continuous data DA on the disc is divided for each block group GRi, and all or some of the sound groups of each block group GRi are repeated. To do so. For example, in the case of 1/2 speed, as shown in FIG. 8B, all the sound groups of each block group GRi are repeatedly reproduced twice. In this case, since discontinuity occurs between the same block groups that repeat, that portion is subjected to overlap processing different from that during normal reproduction.

For example, if the speed is reduced by 25%, every two block groups GR1, GR4, GR7, ... Are extracted from the data DA and each block group GR1, as shown in FIG. 8C. Each of GR4, GR7, ... Is repeated four times. That is, in the second embodiment, the block group GRi to be extracted is extracted according to the low speed.
And the number of repetitions thereof can be combined to achieve an arbitrary low speed.

According to the above, the discontinuity of the MDCT window occurs in block group units, that is, every four or more continuous sound groups, as in the above-described embodiment. Therefore, the number of noises generated is smaller than that in the case of performing variable speed reproduction in units of one sound group, and the noises can be driven to a frequency region where human sensitivity is low.

[Third Embodiment of Variable Speed Reproduction] This third embodiment is a development of the second embodiment. In this example, the target size of the block group GRi extracted from the continuous data string DA is determined in advance, but from the target size,
For example, it is possible to increase or decrease the number of sound groups of 1 to 3 and, as a whole, to enable variable speed reproduction according to a target reproduction speed.

That is, although the target number of sound groups in one block group GRi is 5, for example, 6
When a block group GRi consisting of 4 sound groups is generated, control is performed so as to generate a block group consisting of 4 sound groups as the next or subsequent block group, and as a whole, all 5 The variable speed reproduction is performed exactly as in the case where the blocks are composed of sound groups.

By allowing the size of one block group GRi to vary as described above, the following variable speed reproduction can be implemented in the case of the third embodiment.

As described above, the MD of the sound group
The CT processing has a short mode and a long mode.
Since the short mode is executed in a portion where the sound changes drastically, if the processing for connecting the temporal discontinuity is performed in this short mode portion, the noise generated in that portion is inconspicuous.

Therefore, in the third embodiment, at least the sound groups constituting one block group are arranged so that the sound groups at both ends or one end of the block group are as short as possible. Is generated.

This means that when a block group is generated, the sound groups around it are searched, and if there is a short mode, the block group is generated so that it is at the end. If there is no short mode sound group, it can be realized by generating a block group with a prescribed target number of sound groups.

[Fourth Embodiment of Variable Speed Reproduction] This fourth embodiment is also a development of the third embodiment.

The method of this example is a method in which not only the number of sound groups that make up one block group but also the method of extracting block groups to be extracted in a scattered manner is variable.

That is, for example, one block group is composed of four sound groups, but when it is extracted in a scattered manner, it is a method of extracting so that the short mode sound groups come to the ends as much as possible. In this case, one to three degrees of freedom are set for the number of sound groups forming one block group as in the third embodiment, and if the fifth group is a short mode sound group, for example. If there is one, the block group is composed of 5 sound groups, and another block group consisting of 3 sound groups is generated so that the whole of the blocks is combined.

In the case of this example, there is a high possibility that temporal discontinuities will be connected in the short mode sound group, and it is possible to further reduce the noise audibly.

In the above example, the compression method is MDCT. However, if the compression method is audio data compression processing for performing overlap processing between processing unit blocks, MD is used.
The present invention is applicable not only to CT.

[0101]

As described above, according to the present invention, audio data blocks of a predetermined time length are compressed, and the data blocks of the predetermined time length are overlapped. When reproducing audio data recorded by a compression method that enhances the continuity of the voice waveform at a speed faster or slower than the normal speed without changing the pitch, the number of appearances of noise can be reduced and the frequency of the noise can be reduced. Since human sense of hearing can have low sensitivity, a good reproduced sound can be obtained.

[Brief description of drawings]

FIG. 1 is a diagram for explaining an embodiment of a variable speed reproduction method of audio data according to the present invention.

FIG. 2 is a diagram for explaining an embodiment of a variable speed reproduction method of audio data according to the present invention.

FIG. 3 is a block diagram of an example of a disk device to which the method according to the present invention is applied.

FIG. 4 is a diagram for explaining a compression method of audio data used in the disc device of FIG.

FIG. 5 is a block diagram of a processing portion for realizing an embodiment of a variable speed reproduction method according to the present invention.

FIG. 6 is a process flow chart of an embodiment of a variable speed reproduction method according to the present invention.

FIG. 7 is a diagram for explaining another embodiment of the variable speed reproduction method of audio data according to the present invention.

FIG. 8 is a diagram for explaining another embodiment of the variable speed reproduction method of audio data according to the present invention.

[Explanation of symbols]

 20 system controller 33 audio compression encoding / decoding circuit 50 spectrum signal restoration processing unit 51 to 53 inverse MDCT processing unit for normal reproduction 54 to 57 inverse MDCT processing unit for variable speed reproduction 61, 63 band synthesis filter

Claims (4)

[Claims] 1. A compression method for compressing audio data blocks of a predetermined time length, wherein the data blocks of the predetermined time length are overlapped so that the audio waveform is continuously recorded. When reproducing the audio data, a plurality of minimum decodable audio data blocks are continuous from the recorded audio data that are temporally continuous, and when the plurality of blocks are repeated as a unit, the A block group with a number of blocks whose repetition frequency is low for human auditory perception is extracted repeatedly by thinning out data in the middle, and a block group consisting of a plurality of extracted continuous blocks is extracted. , Data decompression processing and connection of the overlapping part By performing processing, variable speed reproduction method of the audio data to perform high-speed playback than the normal playback speed. 2. The audio data is data compression that combines an orthogonal transformation and a window function, and the orthogonal transformation and the connection processing are performed for each audio data of the predetermined time length. Mode, the orthogonal transformation, and the connection processing are recorded by a compression recording method having a second mode performed for each data of the predetermined time length or less, and the connection between the block groups is performed. If possible, the process is performed in the second
2. The variable speed reproduction method for audio data according to claim 1, wherein the reproduction is performed in a portion recorded in the mode. 3. A compression method for compressing audio data blocks of a predetermined time length, wherein the data blocks of the predetermined time length are overlapped so as to be recorded by a compression method for enhancing the continuity of a voice waveform. When playing back audio data, the minimum decodable audio data block is a plurality of consecutive blocks, and when this block is repeated as a unit, the repetition frequency is human auditory perception sensitivity. For a group of blocks having a low number of blocks, the same one is repeatedly reproduced, and it is regarded as continuous in time. A variable speed playback method for audio data that is played at low speed. 4. Audio data is compressed for each block of a predetermined time length, and the data blocks for each predetermined time length are overlapped so that the continuity of a voice waveform is enhanced and recorded. When reproducing the audio data, a plurality of audio data blocks of the smallest unit that can be decoded are consecutive recorded audio data that are temporally continuous, and when the plurality of blocks are repeated as a unit, the Repetition frequency is human hearing
Divide each block into blocks with low sensitivity, and partially overlap the blocks before and after the block, and set the overlap amount to the playback speed. A variable speed reproduction method for audio data, which is variable according to the speed of the normal reproduction, and a time space is provided between the block groups so that the reproduction speed is lower than that of the normal reproduction.