JP2001339311A - Audio signal compression circuit and expansion circuit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an audio signal compression circuit with a more enhanced compression ratio than that of a conventional circuit and to provide an expansion circuit that restores the compressed signal to the original signal. SOLUTION: Filter banks 24L, 24R divide respectively received digital audio signals into 32 sub-band signals with an equal bandwidth. A sbutractor 26 subtracts a low frequency sub-band signal corresponding to the filter bank 24R from each low frequency sub-band signal of the filter bank 24L. Furthermore, an adder 25 sums each high frequency sub-band signal of the filter bank 24L and a high frequency sub-band signal corresponding to the filter bank 24R. A quantization circuit 31 quantizes and outputs a received signal. A bit stream generating circuit 32 generates a bit stream from the output of a quantization circuit 31 and provides the output of it. An expansion circuit 22 restores bit stream data outputted from a compression circuit 21 into original PCM audio data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an audio signal compression circuit for compressing a digital audio signal based on the MPEG / Audio standard and a decompression circuit for restoring the compressed signal.

[0002]

2. Description of the Related Art Conventionally, MPEG (Moving Picture Coding Experts Group
p) / Audio standard is known. FIG. 3 shows a configuration example of a stereo signal compression circuit 1 and a decompression circuit 2 according to this standard. In this figure, Lin and Rin are PCM (Pu
These are digital audio signals of a Left channel and a Right channel by the lse code modulation, and are supplied to the sub-band analysis filter banks 4L and 4R, respectively. Subband analysis filter bank 4L, 4R
Are input digital audio signals Lin,
Rin is divided into 32 sub-band signals with equal bandwidth. In the figure, only four bands are shown. Each of the quantization units 5L and 5R quantizes the subband signal with the number of quantization bits set for each subband.
The bit stream generation circuit 6 adds bit allocation information and a scale factor for each sub-band to the quantized sub-band samples, adds a header, and creates and outputs a bit stream.

Next, in the decompression circuit 2, a bit stream decomposition circuit 7 extracts bit allocation information and a scale factor from the bit stream and, based on the extracted bit allocation information, extracts Left and Rig from the bit stream.
ht 32 sub-band samples are read and output to the inverse quantization circuits 8L and 8R. Each of the inverse quantization circuits 8L and 8R performs inverse quantization for each subband sample, and further multiplies the result by the scale factor output from the bitstream decomposition circuit and outputs the result. Each of the sub-band synthesis filter banks 9L and 9R synthesizes the sub-quantized sub-band signal and returns the original PCM digital audio signal to output.

On the other hand, as a compression circuit based on MPEG / Audio designed to further increase the compression efficiency than the compression circuit 1 described above, a circuit called a joint stereo is known. FIG. 4 shows a compression circuit 11 using this joint stereo.
FIG. 3 is a circuit diagram showing a configuration of an expansion circuit 12. In this figure, digital audio signals Lin and Rin are divided into 32 sub-band signals by sub-band filter banks 14L and 14R, respectively. The low-frequency sub-band signals of the respective sub-band signals are output to the quantization circuits 15L and 15R as they are, while the high-frequency sub-band signals are added by the adder 15 for each corresponding sub-band. Is output to the quantization circuit 15L. Then, the sub-band signals quantized by the quantization circuits 15L and 15R are output to the bit stream generation circuit 1
6 to generate and output a bit stream.

In the decompression circuit 12, each subband sample output from the bit stream decomposition circuit 17 is inversely quantized by inverse quantization circuits 18L and 18R. The low-frequency subband signals output from the inverse quantization circuits 18L and 18R are output to the subband synthesis filter banks 19L and 19R, respectively, and the high-frequency subband output from the inverse quantization circuit 18L is output. The signal is output to both the sub-band synthesis filter banks 19L and 19R. Subband synthesis filter bank 9
L and 9R respectively combine the dequantized sub-band signals to return to the original PCM digital audio signal and output.

[0006]

The above-mentioned compression circuit 11
Since the low-frequency component is compressed in stereo and the high-frequency component is compressed in monaural, the compression ratio can be further increased as compared with the circuit shown in FIG. 3, which is particularly suitable for compressing stereo signals. However, since the compression circuit 11 also directly compresses the left and right signal components for the low frequency components, there is a problem in that the compression efficiency is still poor in that respect. The present invention has been made in view of such circumstances, and its purpose is to
An object of the present invention is to provide an audio signal compression circuit capable of further increasing the compression ratio as compared with a conventional one, and an expansion circuit for restoring a compressed signal.

[0007]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problem, and the invention according to the first aspect is characterized in that the first and second channel audio signals to be inputted are divided into a plurality of sub-banks by a filter bank. A first channel subband signal output from the filter bank in an audio signal compression circuit which divides the signal into band signals, quantizes the subband signal by quantization means, and further forms and outputs the data in a bit stream state And the second
The lower frequency side of the channel subband signal, and
An audio signal compression circuit comprising a subtraction circuit for calculating a difference between corresponding subband signals and outputting the calculation result to the quantization means.

According to a second aspect of the present invention, there is provided a filter bank for dividing input first and second channel audio signals into a plurality of subband signals, and a first channel output from the filter bank. Subtraction means for calculating a difference between a sub-band signal and a corresponding sub-band signal on the low frequency side of the sub-band signal and the second-channel sub-band signal; and a first-channel sub-band signal output from the filter bank Addition means for calculating the sum of the sub-band signals on the high frequency side of the second channel sub-band signal, and quantization means for quantizing each output of the subtraction means and the addition means, Means for forming an output of said quantization means into data in a bit stream state.

According to a third aspect of the present invention, the input bit stream data is decomposed into individual sub-band signals, dequantized, and then sub-band synthesized by a filter bank to form the original first channel. An audio signal and an audio signal decompression circuit for returning to the second channel audio signal; a low-frequency side corresponding signal of the first channel signal and the second channel signal obtained by the inverse quantization; The audio signal decompression circuit is provided with subtraction means for calculating a difference between the two and outputting the calculation result to the filter bank.

According to a fourth aspect of the present invention, there is provided a bit stream decomposing means for decomposing input bit stream state data into individual sub-band signals, and an inverse quantization means for de-quantizing the sub-band signals. And subtraction means for calculating a difference between a corresponding signal on a low frequency side of the first channel signal and the second channel signal obtained by the inverse quantization, and a second signal obtained by the inverse quantization. A first filter bank for sub-band synthesizing the one-channel signal; and a second filter bank for sub-band synthesizing the output of the subtracting means and the signal on the high frequency side of the first channel signal. An audio signal decompression circuit.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a compression circuit 21 according to the embodiment and a decompression circuit 22 for decompressing data compressed by the compression circuit 21. FIG. 2 is a block diagram showing details of the compression circuit 21. is there. In FIG. 2, each of the PCM digital audio signals Lin and Rin is processed into blocks (called frames) for each predetermined number of samples. Each of the sub-band analysis filter banks 24L and 24R has 32 band-pass filters and divides an input signal into 32 sub-band signals having an equal bandwidth. In this case, each subband signal is downsampled to a 1/32 sampling frequency. Then, the sub-band signal on the lower frequency side of the sub-band signals output from the sub-band analysis filter bank 24L is output to the scale factor extraction / normalization circuit 27 as it is, while being output from the sub-band analysis filter bank 24R. The sub-band signal on the low frequency side of the sub-band signals to be
6 is output. The subtracter 26 subtracts the corresponding low-frequency sub-band signal of the sub-band analysis filter bank 24R from each low-frequency sub-band signal of the sub-band analysis filter bank 24L, and uses the subtraction result as a scale factor extraction / normalization circuit 27. Output to Further, each high-frequency sub-band signal of the sub-band analysis filter bank 24L is added to the corresponding high-frequency side sub-band signal of the sub-band analysis filter bank 24R in the addition circuit 25, and the addition result is used as a scale factor extraction / normalization circuit. 27.

Scale factor extraction / normalization circuit 27
Detects a sample having the maximum absolute value for each subband signal in one frame. The value converted to logarithm and quantized is called a scale factor. And
Divide each subband signal by this scale factor and normalize their values to within ± 1.

On the other hand, the psychological auditory analysis unit 29 calculates a frequency spectrum by FFT (Fast Fourier Transform), calculates a masking threshold value for each subband, that is, an allowable quantization noise power based on the calculated frequency spectrum, and outputs the calculated masking threshold value. I do. The bit allocation unit 30 outputs the output of the psychological hearing analysis unit 29,
The number of quantization bits for each subband is determined by an iterative loop process under the limitation of the number of bits available in one frame determined by the bit rate. The quantization circuit 31 quantizes the subband signal output from the scale factor extraction / normalization circuit 27 with the number of quantization bits set for each subband. The bit stream generation circuit 32 shown in FIG.
Multiplexes the quantized sub-band samples, bit allocation information and scale factor for each sub-band, attaches a header to the multiplexed bit stream, and outputs a bit stream. In FIG. 1, the scale factor extraction / normalization circuit 27, the psychological auditory analysis unit 2
9, the description of the bit allocation unit 30 is omitted.

Further, in the decompression circuit 22 shown in FIG.
Each sub-band sample output from the bit stream decomposition circuit 34 is inversely quantized by the inverse quantization circuits 35L and 35R. The low-frequency subband signals output from the inverse quantization circuit 35L are output to the subband synthesis filter bank 36L, output to the subtractor 36, and output from the inverse quantization circuit 35L. Side sub-band signals are output to both sub-band synthesis filter banks 37L and 37R, respectively. The subtractor 36 subtracts the corresponding sub-band signal of the inverse quantization circuit 35R from the low-frequency side sub-band signal of the inverse quantization circuit 35L, and outputs the result to the sub-band synthesis filter bank 37.
Output to R. Sub-band synthesis filter bank 37L,
Each of the 37Rs combines the inversely quantized sub-band signals, returns the original PCM digital audio signal, and outputs it.

As described above, according to the above embodiment, the sub-band signal on the low frequency side is quantized in the form of (LR). Here, in data compression by MPEG / Audio, the smaller the data, the smaller the compressed data. Therefore, when the stereo signal is (LR), the volume is reduced because there is no monaural component, thereby enabling efficient compression. According to the above embodiment,
One of the low-frequency subband signals of the two channels is treated as L and the other is treated as (LR).
+ R) / 2 and the other may be treated as (LR) / 2.

In the above embodiment, the high-frequency side is (L + R), but this is not the case.
As shown in L, R and R may be used as they are. Also, (L-
The number of subband signals to be R) is arbitrary.

[0017]

As described above, according to the present invention, the difference between the first sub-band signal and the second sub-band signal on the low frequency side and the corresponding sub-band signal is determined. Since the subtraction circuit for performing the operation and outputting the operation result to the quantization means is provided, the data compression efficiency can be improved more than the conventional circuit. Further, when an addition means for calculating the sum of the corresponding subband signals on the high frequency side of the first channel subband signal and the second channel subband signal output from the filter bank is provided, Further, the data compression efficiency can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of the present invention.

FIG. 2 is a block diagram showing details of a compression circuit 21 in FIG.

FIG. 3 is a block diagram showing a configuration example of a conventional audio signal compression circuit and decompression circuit 2 according to the MPEG / Audio standard.

FIG. 4 is a block diagram showing a configuration example of another audio signal compression circuit and expansion circuit 2 based on the conventional MPEG / Audio standard.

[Explanation of symbols]

Reference numeral 21: compression circuit, 22: expansion circuit, 24L, 24R: sub-band analysis filter bank, 25: adder, 26: subtractor, 31L, 31R: quantization circuit, 32: bit stream generation circuit, 34: bit stream decomposition Circuit, 3
5L, 35R: inverse quantization circuit, 36: subtractor, 37L,
37R: Sub-band synthesis filter bank.

Claims (4)

[Claims] An input first and second channel audio signal is divided into a plurality of sub-band signals by a filter bank, and the sub-band signals are quantized by a quantizing means. An audio signal compression circuit for forming and outputting a difference between a first sub-band signal and a second sub-band signal output from the filter bank on a low frequency side and a corresponding sub-band signal; , And a subtraction circuit for outputting the operation result to the quantization means is provided. 2. A filter bank for dividing input first and second channel audio signals into a plurality of sub-band signals, respectively, a first channel sub-band signal and a second channel output from the filter bank. Subtraction means for calculating a difference between the corresponding subband signals on the low frequency side of the subband signals; and a first channel subband signal and a second channel subband signal output from the filter bank. An adding means for calculating the sum of the corresponding subband signals on the high frequency side, a quantizing means for quantizing each output of the subtracting means and the adding means, and a bit stream for outputting the output of the quantizing means. Means for forming state data. An audio signal compression circuit, comprising: 3. An input bit stream state data is decomposed into individual sub-band signals, dequantized, and then sub-band synthesized by a filter bank to obtain original first channel audio signals and second channel audio signals.
In an audio signal decompression circuit for returning to an audio signal, a difference between a signal on a low frequency side and a corresponding signal of the first channel signal and the second channel signal obtained by the inverse quantization is calculated, and the calculation is performed. An audio signal decompression circuit comprising subtraction means for outputting a result to the filter bank. 4. A bit stream decomposing means for decomposing input bit stream state data into individual sub-band signals; an inverse quantization means for de-quantizing the sub-band signals; Subtracting means for calculating a difference between a corresponding signal on the low frequency side of the first channel signal and the second channel signal, and a subband combining means for subband combining the first channel signal obtained by the inverse quantization. 1. An audio signal decompression circuit, comprising: a first filter bank; and a second filter bank for sub-band combining an output of the subtracting means and a high-frequency side signal of the first channel signal.