JPH03265309A - Data compression coding device - Google Patents

Abstract

PURPOSE:To cancel an error caused when a data is decoded and to reproduce the data from an optional point on the way of the data by arranging a straight PCM block in the mode outputting an input signal directly forcibly at an interval of a prescribed number of blocks. CONSTITUTION:An audio PCM signal is stored in a buffer 3 via a switch 2 and fed to a 0-th order filter 5, a 1st order filter 6, a 2nd order filter 7 and a counter 8 in a prediction device 4. The encode filter 6 is configurated known as a differential filter, and an audio PCM signal of one block fed to the 0-th order filter 5 is not subject to differential processing but fed to a range detector 9 as a straight PCM signal. Thus, a DC error component caused by propagation of an error caused at decoding is eliminated and the reproduction from almost an optional point after lapse of a prescribed time is attained.

Description

[Detailed description of the invention]

[Industrial Application Field] The present invention is applicable to, for example, DSP (digital signal processor)
The present invention relates to a data compression encoding device for PCM signals, which is suitable for use in applications such as the present invention. [Summary of the Invention] The present invention provides the highest compression rate among a plurality of modes in which an input signal is divided into blocks of fixed samples and each block is outputted through a plurality of filters, including a mode in which the input signal is directly output. In a data compression encoding device that selects a mode in which an output signal having This is a data compression encoding device that eliminates errors that occur when decoding data and can reproduce data from almost any point in the middle, and also forcibly outputs the input signal directly. In order to arrange the straight PCM blocks in the mode that outputs the input signal directly, the weighting value for mode selection is set to a predetermined value for each of the fixed number of blocks, so that the mode that directly outputs the input signal is forcibly selected. This is a data compression encoding device. (Prior Art) In general, as a method for reducing the transmission bit rate by compressing the number of bits of a supplied PCM signal, for example, a filter that can obtain the highest compression rate is prepared in advance for each block of multiple samples. A so-called filter selection type bit compression encoding system is known in which a filter is selected from among a plurality of filters. The input signal for each block is supplied to each block.The input signal for each block is supplied to the plurality of filters described above, such as a zero-order filter that outputs straight PCM,
The signal is supplied to three filters, a first-order difference filter that outputs a first-order difference and a second-order difference filter that outputs a second-order difference. Then, the maximum absolute value within the block is detected for each filter, and the block data passed through the filter that has the minimum maximum absolute value within the block is selected, requantized from, for example, 16 bits to 4 bits, and output. . In addition, during the above-mentioned requantization, a so-called noise shaving process is performed in which the difference between the manual input and the output of the quantizer is returned to the input side of the quantizer and added to the data newly supplied to the quantizer. It is being done. At the time of the above output, filter information indicating the filter through which the data of the block to be recorded passed, range information indicating the range of the data, etc. are also output together with the above data, and the decoder side performs processing according to these information. The data is played back. [Invention tries to solve it! 18) However, as mentioned above, the filter selection type bit compression encoding system outputs the primary or secondary difference value from the previous data based on the first data in the block, for example, from an arbitrary point. When data is reproduced, if the data reproduced at the arbitrary point is data that has passed through a primary or secondary filter, an error will occur in the reproduced data. When a -degree error occurs in this manner, the error is propagated to subsequent data and a DC error component is generated, making it impossible to obtain good reproduced audio. Therefore, when considering the error margin, there are many problems in using a filter with a high prediction gain. In addition, in order to eliminate this error, Stray 1-P
It is sufficient if CM data is selected, but the straight PCM
It is impossible to predict when data will be selected. In addition, in order to enable playback from any point mentioned above, there is a method of storing the reference value for each block in a storage medium such as a memory, but if such a memory is provided, the decoder side This increases the burden on the hardware, which is particularly undesirable when converting to an IC for the purpose of lowering the price. The present invention has been made in view of the above-mentioned problems, and it eliminates the DC error component caused by the propagation of the above-mentioned error, and allows almost any arbitrary An object of the present invention is to provide a data compression encoding device that can supply encoded data that enables data reproduction from a point. [Means for Solving the Problems] The present invention blocks an input signal for each fixed sample,
For each block, among the multiple modes in which the input signal is output via multiple filters, including the mode in which the input signal is directly output,
In a data compression encoding device that selects a mode that provides an output signal with the highest compression ratio, straight PCM blocks in a mode that directly outputs the input signal are forcibly arranged at intervals of a fixed number of blocks. The above-mentioned t! l! The above-mentioned method is characterized in that the mode for directly outputting the input signal is forcibly selected by setting the weighting value for mode selection to a predetermined value for each of the predetermined number of blocks. Solve problems. (Operation) The data compression encoding device according to the present invention divides an input signal into blocks for each fixed sample, and outputs the input signal via a plurality of filters for each block, including a mode in which the input signal is directly output. Among these, in a data compression encoding device that selects a mode that provides an output signal with the highest compression rate, there is a straight P mode that forcibly outputs the input signal directly at intervals of a fixed number of blocks.
By arranging CM blocks, it is possible to eliminate the DC error component caused by the propagation of errors generated during decoding, and it is possible to reproduce from almost any point after a certain period of time has elapsed. . [ Embodiments] Hereinafter, embodiments of the data compression encoding device according to the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram showing in block form each function of an embodiment of a data compression encoding device according to the present invention. In FIG. 1, audio data generation block 1
can stop reading at any point and can instantly play back from the stopped point. For example, C
It can be constructed using a device that generates digital data, such as a D (compact disc) player or a DAT (digital audio tape recorder), and a memory. Reading from this audio data generation block l is controlled by a control signal from a read control circuit 22, and for example, l samples are set to 16 bits (I word), and 1
An audio PCM signal in which one block is made up of 6 samples is generated, and the audio PCM signal is stored in a buffer 3 via a switch 2, which will be explained later, and is also stored in a buffer 3 through a 0th order filter 5 and a 1st order filter 6 in a predictor 4. .Supplied to a secondary filter 7 and a counter 8. The above encoding filter has a so-called differential filter configuration,
The one block of audio PCM signals supplied to the zero-order filter 5 is supplied as is to the range detector 9 as a so-called straight PCM signal without being differentiated. In this range detector 9, the maximum absolute value within one block of the supplied straight PCM signal is detected, and the maximum absolute value within the block is sent to the comparison (minimum value detection) block 14, and all data of the block is sent to the selector. 15. Next, the l block of audio PCM signals supplied to the first-order filter 6 is subjected to a first-order difference within the block, and this first-order difference data is supplied to the range detector 10. The range detector 10 detects the maximum absolute value within a block from among the primary difference data of one block. The maximum absolute value within the block of this primary difference data is supplied to the weighting unit 12, and the above-mentioned one block of primary difference data is supplied to the selector 15. Then, the maximum absolute value within the block of the primary difference data supplied to the weighting unit I2 is multiplied by a predetermined coefficient (for example, 1.5).
The data are so-called weighted and supplied to a comparison block 14. Next, the l block of audio PCM signals supplied to the secondary filter 7 are supplied to the range detector 11 after a secondary difference within the block is taken. The range detector 11 detects the maximum absolute value within the block from among the supplied one block of secondary difference data. The maximum absolute value within the block of this secondary difference data is supplied to the weighting unit 13, and the secondary difference data of the l block is supplied to the selector 1.
5. The maximum absolute value within the block of the secondary difference data supplied to the weighting unit 13 is multiplied by a predetermined coefficient (for example, 1.5), so-called weighting, and then supplied to the comparison block 14. The comparison block 14 is the straight PCM supplied above.
The maximum absolute value within a block of data, the maximum absolute value within a block of weighted primary difference data, and the maximum absolute value within a block of weighted secondary difference data are compared to find the one with the smallest value. Then, the range of the detected minimum value is output as range information, and it is also detected which filter among the three filters the minimum value has passed through, and this is output as filter information. The range information is supplied to the quantizer 18 and also to the multiplexer 17, and the filter information is supplied to the selector 1.
5 and noise shaving times! 19 and to multiplexer I7. The selector 15 selects the audio PC through the filter specified by the filter information from the comparison block 14.
Select the M signal by, for example, switching a switch,
This selected one block of audio PCM signals is supplied to the quantizer 18. In this way, the selector 15 selects the optimal one from among the audio PCM signals that have passed through the three filters as i! The counter 8 counts the number of blocks of the supplied audio PCM signal, and when it reaches a predetermined count (for example, several blocks to several tens of blocks), the counter 8 The weighter provides a count arrival signal to unit 1.2.13. When each of the weighting units 12 and 13 is supplied with the above-mentioned count arrival signal, the weighting unit 12 and 13, for example, which is 1.5, makes the coefficient sufficiently large (eg, 100, etc.). As mentioned above, since the selector 15 selects the block having the smallest intra-block maximum absolute value from among the three supplied audio PCM signals, it necessarily selects a block of straight PCM signals that have passed through the zero-order filter 5. is selected, and even in the worst case, the block of the straight PCM signal can be forcibly placed after counting the predetermined number of blocks. Here, the above constant 1-71 interval (constant block number count) is N, and the number of samples for 1 bronc is n samples (
For example, in CDI, n=28. In APU, if n-16), it is guaranteed that the 0th order filter 9 will be selected every nN samples, and when reproducing BRR encoded data from any point, at most n N t 5
A complete reproduced sound can be obtained after liecl D s-sampling interval 1 sec+). Specifically, f
, =32 kllz, if one block has 16 samples, the time required for one block is approximately 0.5

[□acl
Therefore, even if straight PCM signals are arranged every several tens of blocks, they can be arranged at intervals of about 20 to 30+□ael. This interval is not perceptible to human hearing, so it does not pose a problem. Although some errors will occur until the block of the straight PCM signal is placed, the data may be masked from the start to the n N L glsacl. 16 selected in this way and supplied to the quantizer 18
For the bit audio PCM signal, depending on the range information supplied from the prediction range adaptation circuit 4, for example, 4 bits are extracted, counting from the first bit of 16 bits, starting from the upper bit. The signal is subjected to so-called requantization and supplied to the multiplexer 17 via the noise shaving circuit 19. The noise shaving circuit 19 converts the so-called quantization error, which is an error between the human power and the output of the quantizer 18, into the prediction range adaptation circuit 14.
It is fed back to the quantizer 18 according to the filter information from the quantizer 18. That is, the data of the lower bits after 4th bin 1 extracted during the re-quantization is processed by the quantizer 18.
The data is returned to the human power of the quantizer 18 and added to the data newly supplied to the quantizer 18 . Here, as mentioned above, the quantizer 1B outputs, for example, 4 bits, counting from the first bit of 16 bits, starting from the most significant bits, based on the range information.
A carry occurs when the quantization error fed back from noise shaving and new data are added, and "1" is transferred to a bit higher than the extraction bit specified by the range information, a so-called overflow. may occur. If this overflow occurs, there is a problem such as a direct current error component occurring during reproduction and good reproduction cannot be achieved.To prevent this, the above weighting is usually performed in the first-order filter and The maximum absolute value in the block through the secondary filter is weighted by multiplying it by a coefficient of, for example, 1.5, and data is extracted from a focus point that is, for example, 1 bit higher than the bit that should originally be extracted. However, even if the above weighting is performed, overflow may occur. Therefore, when the overflow occurs, the noise shaving circuit 19 switches the switch 2. The buffer 3, the mask circuit 21, and the readout control circuit 22 are supplied with overflow information including the overflow amount and the current filter number (0th order, 1st order or 2nd order filter), and the output of the audio PCM signal caused by the overflow is temporarily stopped. Stop. When this overflow information is supplied, the multiplexer 17 stops outputting data, the read control circuit 22 stops reading data from the audio data generation block l, and the switch 2 connects the selection terminal 2c to the selected terminal 2.
A is switched to the selected terminal 2b, and the reception of data from the audio data generation block 1 is stopped. As mentioned above, the buffer 3 stores audio data for several blocks, and the buffer 3 uses the overflow information to perform a so-called retry process in which the audio PCM signal of the block where the overflow occurred is supplied again to the predictor 4. will be held. When this retry is started, the mask circuit 21 learns the current filter number from the supplied overflow information, and supplies a coefficient increment signal to the weighting device connected to this filter. The weighting device to which this coefficient increment signal is supplied has a coefficient value of 1.
The coefficient value is incremented, for example, from 5 to 1.6. The re-supplied audio PCM signal of the block that caused the overflow is passed through each filter to each range detector 9°1 as described above.
0.11, the maximum absolute value within each block is detected, and each data is supplied to the selector 15. Among the detected maximum absolute values within the block, the maximum absolute value within the block of the straight PCM signal via the range detector 9 is supplied as is to the comparison block 14, but the primary difference data via the range detector 10 is The maximum absolute value within the block is supplied to a weighting unit 12, and the maximum absolute value within the block of the secondary difference data via the range detector 11 is supplied to a weighting unit 14. As mentioned above, for either of the above weightings that caused the above overflow, the coefficient value of the unit is incremented, and the maximum absolute value in each block is multiplied by the coefficient value and weighted, and the comparison block is 14. The comparison block 14 compares the maximum absolute values in each block including the maximum absolute value in the block re-weighted by this incremented coefficient, detects the one with the minimum value, and calculates the value again based on this detection result. Provide range information to quantizer Rii+8. Based on this range information, the quantizer 18 performs requantization as described above. Then, when the overflow is eliminated by performing this requantization, the data whose overflow has been eliminated by the requantization is output, and the mask circuit 21 performs the incremented weighting based on the coefficient value of the device. Return to coefficient value, switch 2 is i! The terminal 2c is switched to the selected terminal 2a and the acquisition of new data is resumed. Note that this retry is performed until the overflow is resolved, and if an overflow occurs despite the retry, the weighting coefficient value is incremented again. The requantized 4-bit data is thus supplied to the multiplexer 17. The multiplexer 17 outputs the supplied data, the range information, and the filter information. These data and information are taken out via the output terminal 20. The output data taken out from the output terminal 20 is as shown in FIG. 2, for example, for one bronch, and includes 1 byte of header information (parameter information or attached information related to compression, etc.) RF, and 8 bytes of sample data. It is composed of data DAfi to D113. The above header information 1 [IRF includes 4 bits of range information, 2 bits of the above mode selection information or filter selection information, 2 flag information of 1 bit each, information Ll indicating the presence or absence of a loop, and the end of the waveform. It is composed of information El indicating whether it is a block (end prong) or not. As mentioned above, one sample of data is compressed from, for example, 16 bits to 4 bits, and the above data I)Ao~
D0 contains 4-bit data DamH- for 16 samples.
Contains D13L. Then, such data and various information are recorded, for example, on a recording medium or the like, or directly transmitted to the decoding side and reproduced. As is clear from the above explanation, the data compression encoding device of this embodiment blocks the input signal every 16 samples, and selects one of the 0th-order filter, 5th-order filter, 6th-order filter, and 2nd-order filter to achieve the highest compression. When selecting a filter that can obtain the ratio, requantizing it, and transmitting the output signal, the counter 8 counts the number of input blocks, and when it reaches a predetermined count value, the first filters 6 and 2 Weighting 'l'i12.13 provided in the next filter 7
By setting the coefficient to a predetermined large value, it is forced to 0
Since the next filter is selected and a straight PCM signal can be placed, it is possible to reproduce the data from any point, and in particular, the error that has occurred can be forced to the straight PCM signal at the above-mentioned fixed intervals. Since this problem can be resolved by arranging the signals, it is possible to use a filter with a high prediction gain. In addition, in a data compression encoding device that indirectly controls filter selection by changing weighting parameters for each fixed block, the data compression according to the present invention can be achieved by making only slight changes to such a system. It can be changed to an encoding device. Further, in the data compression encoding device of this embodiment, the output and human power are stopped at the time when an overflow is detected in the noise shaving circuit 19, and the weighting is performed in the noise shaving circuit 19.
By incrementing the coefficient value and processing or trying again the data of the block in which overflow has been detected, the deterioration of the S/N ratio during playback is prevented because overflow data is not supplied. be able to. In addition, the coefficient value of the weighting device above is an example of
For example, it is possible to make multiple changes such as setting the base coefficient to 1.4 and incrementing it to 1.6 at the time of trial, and it goes without saying that the coefficient is not limited to the above values. In a data compression encoding device that selects a mode that provides an output signal with the highest compression ratio among multiple modes in which an input signal is output through multiple filters, including a mode in which the input signal is directly output. By arranging straight PCM blocks in a mode that forcibly outputs the above input signal directly at intervals of a fixed number of blocks, the DC error component that occurs due to the accumulation of errors in the encode-decode system is cleared. Reproduction from approximately any point can be made possible. Further, the data compression encoding device according to the present invention sets a weighting value for mode selection to a predetermined value for each of the predetermined number of blocks so that a mode for directly outputting the input signal is forcibly selected. By doing so, the same effects as described above can be obtained. [Effects of the Invention] The data compression encoding device according to the present invention divides an input signal into blocks for each fixed sample, and divides each block into blocks.

[Brief explanation of drawings]

FIG. 1 is a functional block diagram showing each function of the data compression encoding device according to the present invention in block form, and FIG. 2 is a schematic diagram showing the format of output data of the data compression encoding device according to the present invention. . 22. ...Reading control circuit

Claims (2)

[Claims] (1) The input signal is divided into blocks for each fixed sample, and the output signal with the highest compression ratio is selected among multiple modes in which each block is output through multiple filters, including a mode in which the input signal is directly output. A data compression encoding device that selects a mode obtained by the data compression encoding device, characterized in that straight PCM blocks of a mode that directly outputs the input signal are forcibly arranged at intervals of a fixed number of blocks. . (2) Claim (1) characterized in that a mode for directly outputting the input signal is forcibly selected by setting a weighting value for mode selection to a predetermined value for each of the fixed number of blocks. The data compression encoding device described above.