JP2001094432A - Sub-band coding and decoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sub-band coding and decoding method for a digital audio signal that reduces a delay amount while maintaining its compression rate and to enhance decoding reproduction quality. SOLUTION: A sub-band division processing section 101 applies band split filtering to a received PCM audio signal to decimate the signal. A table value introduction section 106 introduces a table value from a scale factor obtained by a scale factor extract section 104 and a sub-band combining and processing section 107 calculates a combined waveform based on the table value of a preceding scale factor. An energy or maximum value calculation section 108 calculates a maximum value or energy of the combined waveform. A bit assignment processing section 110 calculates a sample assignment bit quantity based on a masking level calculated by an psychroacoustic model 109 and a quantization processing section 111 quantizes a sub-band signal. A frame forming section 112 forms a sign bit, a Huffman code and a sample into a bit stream and outputs the result. Since the processing is conducted with a very short frame, the delay quantity is small.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a sub-band encoding / decoding method, and more particularly to a sub-band encoding / decoding method for compressing and expanding a digital audio signal.

[0002]

2. Description of the Related Art MPEG1 audio encoding (ISO / IEC 11172-3) has been standardized as an international standard as a subband encoding / decoding method for compressing and expanding digital audio signals. FIG. 4 shows a basic structure of an encoder 401 for MPEG1 layer1 audio encoding. A conventional MPEG1 layer1 audio encoding method will be described with reference to FIG.

A digital audio signal is transmitted from an encoder 401.
, The sub-band analysis filter bank 402 performs band division filtering and thinning processing. On the other hand, the input digital audio signal is
At 403, frequency analysis is performed, and the psychoacoustic model 405 and the dynamic bit allocation unit 407 adaptively determine the number of allocated bits for the subband samples based on the frequency analysis information. The sub-band analysis filter bank 40
From the output of 2, the scale factor value is calculated by the scale factor extraction unit 404, and the output of the sub-band analysis filter bank is quantized by the linear quantizer 406. Also, the scale factor, the bit allocation information, and the like are encoded by the side information encoding unit 409 as side information. A bit stream is generated by a bit stream forming unit 408 by combining ancillary data with a sample output from the linear quantizer 406 and a side information code output from the side information encoding unit 409.

FIG. 6 shows the structure of a bit stream. The bit stream is composed of a header 601, allocation information 60
2. It is composed of a scale factor 603 and a sample 604. FIG. 5 shows a decoder that receives the bit stream and decodes the audio data and performs MPEG1 layer 1 audio decoding. Referring to FIG. 5, a conventional MPEG1
The layer1 audio decoding method will be described.

[0005] The decoder 501 receives the stream information sent through the communication path or the storage medium, and the bit stream decomposing unit 502 decomposes the stream. Further, a scale factor and bit allocation information are extracted as side information in the side information decoding unit 503, the subband signal is inversely quantized in the inverse quantization unit 504, and inversely quantized in the subband synthesis filter bank 505. The sub-band signal is synthesized with the original digital audio signal and output.

[0006]

However, in the conventional subband encoding / decoding method, buffering of the digital audio signal for frame analysis occurs, and if the amount of buffering is large, real-time performance is required. In systems, there is a problem that non-negligible delay occurs.

SUMMARY OF THE INVENTION The present invention solves the above-mentioned conventional problems. In a method for encoding / decoding a sub-band of a digital audio signal, the delay amount is reduced and the decoding / reproduction quality is improved while maintaining a compression rate. It is an object of the present invention to provide a subband encoding / decoding method.

[0008]

According to the present invention, a table value is derived from a scale factor for each subband obtained from a subband signal obtained by dividing an input digital audio signal into subbands. Calculate sample bit allocation information based on the energy or maximum value of the waveform obtained by combining and filtering the derived table value and the table value derived from the scale factor values of a plurality of frames before the current frame, and A subband encoding method of quantizing a subband signal based on bit allocation information to calculate a sample value is performed by calculating a difference between a scale factor of a current frame and a scale factor of each subband of a previous frame, and calculating an absolute value of the difference. A code consisting of a Huffman code, a sign bit indicating whether the difference is positive or negative, and a sample value. Configured to output a stream.

[0009] With this configuration, the temporal length of the frame is the time corresponding to the number of digital audio signals corresponding to the number of sub-band divisions, and a high-quality digital audio signal with low delay can be decoded. Furthermore, the sample allocation information in the bit stream output from the encoder is excluded, the bit amount allocated to the sample is increased, and the bit allocation amount in consideration of the energy and the maximum value of the thinned waveform portion can be calculated.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to FIGS.

(First Embodiment) In a first embodiment of the present invention, a table value derived from a scale factor for each subband and a table value derived from a previous scale factor value are synthesized and filtered. Calculates the sample value by quantizing the sub-band signal based on the sample bit allocation information calculated based on the energy or the maximum value of the obtained waveform, and calculates the Huffman code of the absolute value of the difference between the scale factors for each sub-band. And a sub-band encoder that outputs a stream consisting of a sign value indicating the sign of the difference and a sample value.

FIG. 1 is a functional block diagram of a sub-band encoder according to the first embodiment of the present invention. FIG.
, The sub-band division processing unit 101 is a unit that performs band division filtering on an input digital audio signal and performs thinning processing, and is configured by M division band filters and M down samplers. . The scale factor extraction unit 104 is a means for extracting a scale factor representing a magnification of a down-sampled sub-band signal for each sub-band. Difference value extraction unit 103
Is a means for calculating a difference value with respect to the scale factor value of the previous frame and taking the absolute value thereof. The sign bit determination unit 102 is a means for obtaining the sign of the difference.
The Huffman encoding unit 105 is a means for Huffman encoding the absolute value of the scale factor difference calculated by the difference value extracting unit 103. The table value deriving unit 106 is means for deriving a table value from a scale factor. The subband synthesis processing unit 107 is a unit that performs subband synthesis processing based on the derived table value and the table value derived from the scale factor before the current frame, and calculates a synthesized waveform. The energy or maximum value calculation unit 108 is means for calculating the maximum value or the energy of the combined waveform in the frame from the combined waveform. The psychoacoustic model 109 is means for calculating a masking level of a quantization error based on psychoacoustics based on the calculated maximum value or energy. The bit allocation processing section 110 is a means for calculating the number of sample allocated bits for each subband. Quantization processing section
Reference numeral 111 denotes means for quantizing the thinned subband signal as sample information based on the number of bits allocated to the sample. The frame forming unit 112 is a means for forming ancillary data, sign bits, Huffman codes, and samples into a stream.

FIG. 2 is a diagram showing a configuration of a stream output from the subband encoder according to the first embodiment of the present invention.

The operation of the subband encoder according to the first embodiment of the present invention configured as described above will be described with reference to FIG.
This will be described with reference to FIG. First, digital audio signals for the number of divisions are input to the sub-band encoder.
If an M-divided filter bank is used, M PCs
The M signal is input. The input signal is subjected to band division filtering in the sub-band division processing unit 101, and is output as a signal decimated every M samples. At this time, the thinned subband signal per frame of each subband filter output is one sample.

The output of one sample of each subband is scaled by a scale factor extracting section 104 to obtain a scale factor. The M scale factors for each subband are input to table value deriving section 106, and a table value determined by the scale factor is derived.
The table value and the table value derived from the scale factor before the current frame and the subband synthesis processing are performed in the subband synthesis processing unit 107 to calculate a synthesized waveform.

In the energy or maximum value calculation unit 108,
The maximum value or energy of the synthesized waveform portion for the newly generated frame is calculated, and the masking level of the quantization error based on the psychoacoustic is calculated by the psychoacoustic model 109. Based on this masking level, the sample allocation bit amount of each subband signal is calculated by the bit allocation processing unit 110, and the decimated subband signal is converted into the sample allocation bit amount by the quantization processing unit 111. Quantized based on The frame forming section 112 shapes the sign bit, the Huffman code, the sample, and possibly the ancillary data into a bit stream and outputs it. In this subband encoding method, the number of bits allocated to the sample is increased and processing is performed in a very short frame, so that the amount of delay is small.

FIG. 2 shows the structure of a stream output from the encoder. The absolute value 201 of the Huffman-encoded scale factor difference for each frame and a sign bit 202 indicating the sign of the scale factor difference for each frame
And the sample information for the subband.

As described above, in the first embodiment of the present invention, the subband encoder combines the table value derived from the scale factor of each subband with the table value derived from the previous scale factor value. The sample value is calculated by quantizing the subband signal based on the sample bit allocation information calculated based on the energy or the maximum value of the waveform obtained by filtering, and the Huffman of the absolute value of the difference of the scale factor difference for each subband is calculated. Since a stream composed of a code, a sign bit indicating the sign of the difference, and a sample value is output, the number of bits allocated to the sample is increased, and subband encoding can be performed at a low bit rate with a low delay.

(Second Embodiment) In a second embodiment of the present invention, a Huffman code of the absolute value of a scale factor difference, a stream composed of an increase / decrease code thereof, and a sample are received, and a table derived from the scale factor is received. This is a sub-band decoder that performs sub-band synthesis on the values and the table values of the previous frame to obtain the energy or the maximum value and allocate bits.

FIG. 3 is a functional block diagram of a sub-band decoder according to the second embodiment of the present invention. In FIG. 3, an ancillary data extraction unit 301 is a means for extracting ancillary data from an input bit stream. The Huffman decoding unit 303 is means for decoding the Huffman code for the subband from the received bit stream and calculating the absolute value of the difference from the scale factor value of the previous frame. The sign bit extracting unit 302 is means for extracting a sign bit indicating the increase or decrease of the scale factor of the previous frame from the stream after extracting the ancillary data and the Huffman code. The scale factor calculation unit 309 is means for calculating the scale factor value of the current frame from the absolute value of the scale factor difference of the previous frame, the sign bit, and the scale factor value of the previous frame. The table value deriving unit 308 is means for calculating a table value from the scale factor value of the current frame. The subband synthesis processing unit 307 is a unit that performs subband synthesis processing based on the calculated table value and the table value before the current frame, and calculates a synthesized waveform. The energy or maximum value calculation unit 306 is means for calculating the maximum value or energy of the composite waveform in a frame section newly generated from the composite waveform. The psychoacoustic model 305 is means for calculating a masking level of a quantization error based on psychoacoustic from the calculated maximum value or energy. The bit allocation processing unit 304 is means for calculating the number of bits allocated to the samples allocated to each subband signal based on the masking level. The sample value extracting section 310 is a means for extracting a sample value from the stream from which the ancillary data, the Huffman code, and the increase / decrease sign bit have been removed, based on the bit allocation information. The inverse quantization unit 311 is a unit that inversely quantizes the subband signal based on the scale factor and the sample value. Subband synthesis processing unit 312
Is means for performing synthesis filtering of the dequantized sub-band signal and outputting a digital audio signal.

The operation of the subband decoder according to the second embodiment of the present invention configured as described above will be described with reference to FIG. First, a bit stream for one frame is input to the subband decoder, and ancillary data is extracted by the ancillary data extraction unit 301. When ancillary data is not included, the ancillary data extraction unit 301 may be omitted. Further, the stream from which the ancillary data has been extracted is sent to the Huffman decoding unit 303, where the absolute value of the difference between the scale factor and the previous frame is calculated. From the stream data after the extraction of the Huffman code, a sign bit indicating an increase or decrease in the scale factor value from the previous frame is extracted by a sign bit extracting unit 302.

Based on the absolute value of the scale factor difference, the sign bit and the scale factor value of the previous frame,
Scale factor calculation section 309 calculates the scale factor of the current frame. In the table value derivation unit 308,
A table value is derived based on the calculated scale factor, and a subband synthesis processing unit 307 calculates a composite waveform from the table values before the current frame. In this waveform, the energy or maximum value of the newly generated waveform for the frame section is calculated by the energy or maximum value calculation unit 306.
Is calculated.

The psychoacoustic model 305 calculates a masking level of a quantization error based on psychoacoustic based on the maximum value or the energy. Bit allocation processing unit 304
Then, based on the information of the psychoacoustic model 305, the bit amount allocated to each subband sample is calculated. The stream data after extracting the ancillary data, the Huffman code, and the increase / decrease sign bit are sampled
, And a sample value for each subband is extracted based on the bit allocation information from the bit allocation processing unit 304.

Based on the scale factor value and the sample value,
The sub-band signal is inversely quantized by the inverse quantization unit 311.
This inversely quantized subband signal is supplied to a subband synthesis processing unit.
At 312, it is combined with the PCM digital data and becomes the output of the decoder.

As described above, in the second embodiment of the present invention, the sub-band decoder receives the Huffman code of the absolute value of the scale factor difference, the stream composed of the increase / decrease code thereof, and the sample, and derives from the scale factor. The sub-band is synthesized by combining the table value obtained from the previous frame and the table value of the previous frame to determine the energy or the maximum value, thereby increasing the number of bits allocated to the sample, and performing sub-band decoding at a low bit rate with low delay. can do.

[0026]

As is apparent from the above description, according to the present invention, a table value is derived from a scale factor for each subband obtained from a subband signal obtained by dividing an input digital audio signal into subbands. The sample bit allocation information is calculated based on the energy or the maximum value of the waveform obtained by combining and filtering the value and the table value derived from the scale factor values of a plurality of frames before the current frame. A sub-band encoding method of quantizing the sub-band signal based on the basis to calculate a sample value, obtains a difference between a scale factor of a current frame and a scale factor of each sub-band of a previous frame, and a Huffman code of an absolute value of the difference, A stream consisting of a sign value indicating the sign of the difference and a sample value is output. Since the configuration and the to, there is no need to allocate bit allocation information in the stream, further the sample allocation information is increased, the effect is obtained that decoding can be reproduced with high quality low delay even at low bit rates.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a subband encoder according to a first embodiment of the present invention;

FIG. 2 is a configuration diagram of an output bit stream of a subband encoder according to the first embodiment of the present invention;

FIG. 3 is a functional block diagram of a subband decoder according to a second embodiment of the present invention;

FIG. 4 is a functional block diagram of a conventional MPEG1 layer1 encoder;

FIG. 5 is a functional block diagram of a conventional MPEG1 layer1 decoder;

FIG. 6 is a configuration diagram of an output bit stream of a conventional MPEG1 layer1 encoder.

[Explanation of symbols]

101 Subband division processing unit 102 Sign bit determination unit 103 Difference value extraction unit 104 Scale factor extraction unit 105 Huffman coding unit 106 Table value derivation unit 107 Subband synthesis processing unit 108 Energy or maximum value calculation unit 109 Auditory psychological model 110 bits Allocation processing unit 111 Quantization processing unit 112 Frame formation unit 201 Huffman-coded absolute value of scale factor for each frame 202 Sign bit 203 indicating sign of scale factor difference for each frame 203 Sample information for subbands 301 Ancillary data Extraction unit 302 Sign bit extraction unit 303 Huffman decoding unit 304 Bit allocation processing unit 305 Psychoacoustic model 306 Energy or maximum value calculation unit 307 Subband synthesis processing unit 308 Table value derivation unit 309 Scale factor calculation unit 310 Sample value extraction unit 311 Inverse Quantization unit 312 Subband synthesis processing unit 401 Encoder 40 2 Subband analysis filter bank 403 FFT unit 404 Scale factor extraction unit 405 Psychological psychology model 406 Linear quantizer 407 Dynamic bit allocation unit 408 Bitstream formation unit 409 Side information encoding unit 501 Decoder 502 Bitstream decomposition unit 503 Side information Decoding 504 Inverse quantization 505 Subband synthesis filter bank 601 Header 602 Allocation information 603 Scale factor 604 samples

Claims (4)

[Claims] 1. A table value is derived from a scale factor for each subband obtained from a subband signal obtained by dividing an input digital audio signal into subbands, and the derived table value and a scale factor of a plurality of frames before a current frame are derived. Calculate sample bit allocation information based on the energy of the waveform obtained by combining and filtering the table value derived from the value, and calculate the sample value by quantizing the sub-band signal based on the sample bit allocation information. In the sub-band encoding method, the difference between the scale factor of the current frame and the scale factor of each sub-band of the previous frame is obtained, the Huffman code of the absolute value of the difference, a sign bit indicating the sign of the difference, and the sample Output a stream composed of values and Band encoding method. 2. A table value is derived from a sub-band scale factor obtained from a sub-band signal obtained by dividing an input digital audio signal into sub-bands. The derived table value and a scale factor of a plurality of frames before a current frame are derived. Sample bit allocation information is calculated based on the maximum value of the waveform obtained by performing synthesis filtering with the table value derived from the value, and the subband signal is quantized based on the sample bit allocation information to obtain a sample value. In the subband encoding method to calculate, the difference between the scale factor of the current frame and the scale factor of each subband of the previous frame is obtained, the Huffman code of the absolute value of the difference, a sign bit indicating the sign of the difference, Outputting a stream composed of sample values Coding method. 3. A stream generated by the sub-band encoding method according to claim 1, wherein an absolute value of a difference between a scale factor value of a previous frame and a scale factor value of a current frame for each sub-band is Huffman-decoded. Ask
A Huffman-decoded value extracts a sign bit indicating whether the difference between the scale factors of the subbands other than 0 is positive or negative, and calculates a scale factor value for each subband of the current frame,
Deriving a table value, calculating sample bit allocation information based on the energy of the waveform obtained by performing synthesis filtering on the table value derived from the scale factor values of a plurality of frames before the current frame, and obtaining sample bit allocation information A sub-band decoding method comprising: extracting a sample from a stream based on a sub-band, dequantizing and obtaining a sub-band signal from a scale factor and a sample, and reproducing a digital audio signal through a sub-band synthesis filter. 4. A stream generated by the sub-band encoding method according to claim 2, wherein an absolute value of a difference between a scale factor value of a previous frame and a scale factor value of a current frame for each sub-band is Huffman-decoded. Ask
A Huffman-decoded value extracts a sign bit indicating whether the difference between the scale factors of the subbands other than 0 is positive or negative, and calculates a scale factor value for each subband of the current frame,
A table value is derived, and sample bit allocation information is calculated based on the maximum value of the waveform obtained by combining and filtering the table value derived from the scale factor value of a plurality of frames before the current frame and the sample bit allocation. A subband decoding method, comprising extracting a sample from a stream based on information, dequantizing and obtaining a subband signal from the scale factor and the sample, and reproducing the digital audio signal through a subband synthesis filter.