JP3342996B2 - Multi-channel audio encoder and encoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently encode the audio signals of multiple channels with a small quantity of bit by APCM-quantizing the multiple channels audio signals of a frequency area through the use of the result of bit allocation in a bit a location unit. SOLUTION: An APCM pre-processing part 120 inputting the stereo audio signals, and the multiple channel audio signals of the frequency area and calculating information on the size of respective frequency identification signals and a multiple channel processor 130 allocating the channels through the use of a correlation among the multiple channels and signals size information are provided. The bit allocation unit 140 allocating the bits to the respective audio signals by using the result of the multiple channel processor 130 and the result of a psychological acoustic part 170 and a quantizer 150 quantizing the multiple channel audio signals in the frequency area by using the bit allocated to the bit allocation unit 140 are provided. Thus, the audio signals of the multiple channels can be transmitted by encoding the audio signals of the multiple channels by using APCM and the correlation among the channels.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multi-channel audio encoder and an encoding method, and more particularly to an APCM (Ad).
active Plus Coded Modula
The present invention relates to an encoder and an encoding method for efficiently encoding a multi-channel audio signal with a small number of bits using the correlation between each channel and each channel.

[0002]

2. Description of the Related Art In conventional encoders and decoders for multi-channel audio, although the number of transmission channels provided is limited, the number of original signal channels tends to increase, so that transmission with sufficient sound quality is performed. It becomes difficult to do. For this reason, some audio codec algorithms may place more emphasis on the directionality of the sound during playback than on the correct transmission of the sound.

The conventional audio codec algorithm has the interchangeability with the existing stereo system, so that the original signal is mixed to create a stereo channel. In this case, when a multi-channel signal is created again by the stereo channel at the time of decoding, the stereo reproduction is performed. Sometimes, inaudible quantization distortion may occur. This can be a serious problem if the transmitted multi-channel signal is further mixed and used.

On the other hand, an audio codec algorithm according to ISO / IEC 13818-3, which is an MPEG-2 audio standard, includes a sub-band filtering process, a matrixing process, a first scale factor calculating process, a first transmission pattern calculating process, a psychoacoustic process. SMR (Signal-to-Masking Ra) by Model
tio) value calculation process, transmission channel allocation process, dynamic crosstalk suppression process, delay compensation value, prediction coefficient and prediction gain value calculation process, prediction selection information calculation process, modified S
MR value calculation process, prediction coefficient quantization process, initial prediction error signal calculation process, second scale factor calculation process, second
A transmission pattern calculation process, a bit allocation process using the corrected SMR value, a first subband sample quantization process,
Dequantizing the sub-band samples, calculating the final prediction error signal using the de-quantized sub-band samples, calculating the third scale factor, calculating the third transmission pattern, quantizing the second sub-band samples, It consists of a bit stream formatting process. The above-mentioned audio codec algorithm is not only very complicated, but also it is necessary to repeat the routine several times, such as repeating the scale factor calculation process and the transmission pattern calculation process three times. There is a disadvantage that the psychoacoustic model can be applied only to the signal.

[0005]

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to solve the above-mentioned problems by efficiently using multi-channel audio signals with a small number of bits by utilizing the correlation between the APCM and each channel. An object of the present invention is to provide a multi-channel audio encoding method for encoding.

It is another object of the present invention to provide an encoder most suitable for implementing the above-mentioned other channel audio encoding method.

[0007]

According to another aspect of the present invention, there is provided a multi-channel audio encoder for encoding an audio signal input to multiple channels. The masking level of each channel calculated for the stereo audio signal and the multi-channel audio signal, the magnitude information of each frequency signal for the stereo audio signal and the multi-channel audio signal in the frequency domain, and the correlation between the channels. A multi-channel processing step of deciding whether or not to transmit the audio signal of each channel, and allocating a channel; and a result of channel allocation in the multi-channel processing step and the masking level. A bit allocation step of allocating bits to The multi-channel audio signal APC in the frequency domain by using the results of the bit allocation in extent
And a quantization step of performing M quantization.

According to another aspect of the present invention, there is provided a multi-channel audio encoder according to the present invention, comprising: a first mixer for receiving a multi-channel audio signal in a time domain to form a stereo audio signal; A second mixer for forming a stereo audio signal by inputting the converted multi-channel audio signal, a time-domain stereo audio signal output from the first mixer, and a time-domain multi-channel audio signal And a psychoacoustic unit for calculating a masking level of each channel, and magnitude information of each frequency-dependent signal with respect to a frequency domain stereo audio signal and a frequency domain multi-channel audio signal output from the second mixer; Enter the masking level calculated by the psychoacoustic unit and A multi-channel processor for deciding whether or not to transmit the audio signal of the channel, and allocating a channel allocation result in the multi-channel processor and the masking level to each of the audio signals; And a bit allocator for allocating bits to the multi-channel audio signal in the frequency domain using the bit allocation result of the bit allocator.
And a quantizer for performing quantization.

[0009]

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

FIG. 1 is a block diagram showing an embodiment of a multi-channel audio encoder according to the present invention.

The configuration of the block diagram shown in FIG. 1 includes a filter 100 for converting a multi-channel audio signal into a frequency-domain signal, and a second multi-channel audio signal output from the filter 100 as a stereo signal. A mixer 110, an APCM preprocessing unit 120 that receives a stereo audio signal in a frequency domain and a multi-channel audio signal output from the second mixer 110 and calculates magnitude information of a signal for each frequency, and a correlation between the multi-channels A multi-channel processor 130 for allocating channels using the relationship and signal size information, and a first mixer 16 for converting a multi-channel audio signal in the time domain into a stereo signal
0, a psycho-acoustic unit 170 that receives a multi-channel audio signal and a stereo signal in the time domain to calculate a masking level of each channel, and uses the result of the multi-channel processor 130 and the result of the psycho-acoustic unit 170 A bit allocator 140 for allocating bits to the audio signal, and a quantizer 150 for quantizing a frequency-domain multi-channel audio signal using the bits allocated by the bit allocator 140
Consisting of

FIG. 2 shows a multi-channel processor 13 shown in FIG.
FIG. 7 is a detailed block diagram illustrating a multi-channel audio signal in the frequency domain for each frequency, comparing the magnitude of the multi-channel audio signal for each frequency, assigning the smallest signal, and assigning a channel; It comprises a correlation coefficient processor 132 which determines whether or not to transmit according to the number of relations and allocates a channel.

FIG. 3 is a detailed block diagram of the quantizer 150 shown in FIG. 1. The first quantizer quantizes a multi-channel audio signal in the frequency domain using the result of bit allocation by the bit allocator 140. 151 and the first quantizer 1
An inverse quantizer 152 for restoring the signal quantized at 51 as a signal before quantization, and an audio signal restored by the inverse quantizer 152 and a multi-channel audio signal in the frequency domain are mixed to form a stereo signal. A second stereo signal generator 153 for quantizing a frequency-domain stereo audio signal output from the stereo signal generator 153 using the bit allocation result of the bit allocator 140;
It comprises a quantizer 154 and a bit stream former 155 for forming a bit stream using the signal quantized by the first quantizer 151 and the signal quantized by the second quantizer 154.

FIG. 4 is a block diagram of the correlation coefficient processor 132 shown in FIG.
4 is a flowchart for explaining the operation of FIG.
Steps 403 to 403 are steps of obtaining a correlation coefficient between signals at the highest frequency among the given multi-channel audio signals, and comparing the correlation coefficients with each other to obtain a channel having the largest correlation coefficient. Stage or 40
Step 6 is a step of determining whether the channel having the largest correlation coefficient is a stereo channel and performing a corresponding process. Steps 407 to 410 are selected when the expected bit generation amount is larger than the available bit amount. This is a process of repeating the above process for all audio signals in the obtained frequency domain.

The operation of the present invention will be described with reference to FIGS.

Referring to FIG. 1, an input multi-channel audio signal is converted into a signal in a frequency domain through a filter 100.

The given time-domain audio signal is mixed by the first mixer 160 in a form suitable for reproduction on an existing stereo system. On the other hand, the audio signal changed to the frequency domain by the filter 100 is also transmitted to the second mixer 11.
The audio signals in the frequency domain corresponding to the stereo system are formed by mixing at 0.

The stereo audio signal in the time domain output from the first mixer 160 and the multi-channel audio signal in the original time domain are input to the psychoacoustic unit 170 and their masking levels are calculated.

The original multi-channel audio signal and stereo audio signal in the frequency domain are combined with the APCM pre-processing unit 12.
0 is inputted to calculate the magnitude information of the audio signal for each frequency required for the APCM.

A output from APCM preprocessing unit 120
The magnitude information of the audio signal for each frequency required for the PCM and the masking level value output from the psychoacoustic unit 170 are input to the multi-channel processor 130, and the correlation between channels and the magnitude of the signal level between channels are input. The signal to be transmitted is distinguished from the signal that does not need to be transmitted according to the difference. This distinction is unnecessary because the channel capacity of the transmission channel is not large enough to transmit both a given multi-channel signal and a newly formed stereo signal.
The operation of the multi-channel processor 130 will be described in more detail with reference to FIG.

In FIG. 2, the channel allocator 131
It is determined whether to transmit the audio signal using the magnitude information between the signals given from the PCM preprocessing unit (120 in FIG. 1), that is, the magnitude of the multi-channel audio signal is compared for each frequency. And determine the smallest signal to be transmitted. On the other hand, there is a case where the signal having the smallest average magnitude is determined as a signal to be transmitted in a certain time section.

The correlation coefficient processor 132 determines whether to transmit an audio signal using the correlation coefficient between channels. This will be described in more detail with reference to the flowchart shown in FIG.

In FIG. 4, at step 400, a frequency-domain stereo signal formed by mixing a frequency-domain multi-channel signal converted from a time-domain multi-channel signal and a frequency-domain multi-channel signal is input. In steps 401 to 403, after selecting a channel having the largest correlation coefficient in the highest frequency region, steps 404 to 40 are performed.
In step 5, when one of the channels is a stereo channel, it is determined that the channel corresponding to the stereo channel is not transmitted except for the size component. In step 406, if two of the original multi-channels have the largest correlation coefficient, the average value of the two channels is put into one channel, and the other channels do not transmit, leaving only the magnitude component.

In steps 407 to 408, the expected bit generation amount is calculated and, if less than the available bit amount, this flowchart is terminated. If it is larger, it is determined whether all audio signals in the selected frequency domain have been processed. (4
09 stage).

As a result of the determination in step 409, if all the signals in the selected frequency domain have been processed, a signal in a frequency domain lower than the highest frequency domain is selected in step 410, and the process returns to step 402. If not, in step 411, the channel having the next larger correlation coefficient is selected and processed in the same manner as described above.

If the amount of data to be generated is large even when all the channels are used by repeating the above-described process, the process is repeated in the frequency domain below the generated data. On this occasion,
The amount of data to be generated can be calculated from the number of bits that causes quantization distortion to occur at a level lower than the masking level of the transmitted signal by a predetermined value.

Using the masking level of each channel calculated by the psychoacoustic unit 170, the processing result of the multi-channel processor 130, and the processing result of the APCM preprocessing unit 120, the bit allocator 140 assigns bits to each audio signal. Assign. That is, bits are not assigned to a signal determined not to be transmitted by the multi-channel processor 130, but bits are assigned only to a signal determined to be transmitted and a stereo signal.

The quantizer 150 uses the result of the bit allocator 140 to apply APC to the output signal of the filter 100.
Perform M quantization and transmit the result together with other information. This will be described in more detail with reference to FIG.
The quantizer 150 first encodes the original other channel signal output from the filter 100 using the allocated bits, and then decodes this signal and mixes it with a signal not to be transmitted to produce a further stereo signal. The stereo signal is encoded using the allocated bits.

[0029]

As described above, in the multi-channel audio encoder and the encoding method according to the present invention, a multi-channel audio signal is encoded using the correlation between the APCM and each channel, so that a small number of audio signals are encoded. Audio signals of many channels can be transmitted through the transmission channel, and can be transmitted in a manner compatible with existing APCM decoders.

It is also possible to decode the data and reproduce it on a stereo system. When a multi-channel audio signal is further formed from the transmitted stereo signal, or when this signal is mixed again to form a stereo signal, the encoding is performed. Distortion becomes inaudible.

[Brief description of the drawings]

FIG. 1 is a block diagram of a multi-channel audio encoder according to an embodiment of the present invention.

FIG. 2 is a detailed block diagram of a multi-channel processor in FIG.

FIG. 3 is a detailed block diagram of a quantizer in FIG. 1;

FIG. 4 is a flowchart for explaining the operation of the correlation coefficient processor in FIG. 2;

Continuation of front page (56) References JP-A-1-318327 (JP, A) JP-A-4-360331 (JP, A) JP-A-5-227039 (JP, A) , A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H03M 7/38 G10L 19/00 H04B 14/04

Claims (10)

(57) [Claims] 1. A first mixer for inputting a multi-channel audio signal existing in a time domain to form a stereo audio signal, and inputting a multi-channel audio signal converted to a frequency domain to form a stereo audio signal. A psycho-acoustic unit for calculating a masking level of each channel by inputting a stereo audio signal in the time domain and a multi-channel audio signal in the time domain output from the first mixer; Inputting the stereo audio signal in the frequency domain output from the second mixer and the magnitude information of each frequency signal for the multi-channel audio signal in the frequency domain, and the masking level calculated from the psychoacoustic unit, and Determine whether to transmit audio signals and assign channels. A multi-channel processor for assigning, a bit allocation unit for inputting a result of channel allocation of the multi-channel processor and the masking level and allocating bits for each audio signal; and a bit allocation in the bit allocator. And a quantizer for APCM-quantizing the frequency-domain multi-channel audio signal using the result of the multi-channel audio signal. 2. The multi-channel processor, comprising: a channel allocator that compares the magnitude of the multi-channel audio signal in the frequency domain for each frequency, determines the signal having the smallest magnitude as a signal to be transmitted, and allocates a channel; 2. The multi-channel system according to claim 1, further comprising a correlation coefficient processor for calculating a correlation coefficient between the respective channels, and determining whether or not to transmit according to the calculated correlation coefficient. Audio encoder. 3. The channel allocator according to claim 2, wherein a signal having the smallest average magnitude for a predetermined time in each frequency band of the multi-channel audio signal is determined as a transmission signal, and a channel is allocated. A multi-channel audio encoder as described. 4. The correlation coefficient processor determines that when a stereo channel is selected as a channel having the largest correlation coefficient in the highest frequency band, a signal corresponding to the stereo channel is not transmitted while leaving only a magnitude component. 3. The multi-channel audio encoder according to claim 2, wherein: 5. The correlation coefficient processor calculates an average value of audio signals transmitted to the two channels when one of the two channels is selected as a channel having the largest correlation coefficient in the highest frequency band. 3. The multi-channel audio encoder according to claim 2, wherein the remaining channel is determined not to be transmitted while leaving only the magnitude component. 6. The correlation coefficient processor selects a channel having the largest correlation coefficient for the highest frequency band, determines whether or not to transmit according to the characteristics of the channel, and determines the highest frequency band. 3. The multi-channel audio encoder according to claim 2, wherein when all the multi-channel audio signals are processed for the band, the processing is repeated for a frequency region below the multi-channel audio signal. 7. The correlation coefficient processor calculates an expected bit generation amount using the masking level, and when the expected bit generation amount is larger than an available bit amount, the correlation coefficient processor calculates the expected bit generation amount with respect to the lower frequency domain. 7. The multi-channel audio encoder according to claim 6, wherein the processing steps are repeated. 8. The first quantizer, wherein the quantizer quantizes the multi-channel audio signal in the frequency domain using a result of bit allocation in the bit allocator; An inverse quantizer for restoring the quantized signal as a signal before quantization; and a stereo for forming a stereo signal by mixing the audio signal restored by the inverse quantizer with the multi-channel audio signal in the frequency domain. A signal quantizer; a second quantizer for quantizing a frequency-domain stereo audio signal output from the stereo signal generator using a result of the bit allocation in the bit allocator; and the first quantizer. And a bit stream generator for forming a bit stream using the signal quantized in step (a) and the signal quantized in the second quantizer. Multi-channel audio encoder according. 9. A multi-channel audio encoder for encoding an audio signal input to a multi-channel, comprising: a masking level of each channel calculated according to the stereo audio signal in the time domain and the multi-channel audio signal; The size information of each frequency signal for the stereo audio signal and the multi-channel audio signal in the frequency domain, and whether or not to transmit the audio signal of each channel is determined based on the correlation between the channels, and the channel is determined. A multi-channel processing step of allocating, a bit allocation step of allocating bits to each audio signal according to a result of channel allocation in the multi-channel processing step and the masking level, and a result of bit allocation in the bit allocation step. The multi-channel audio signal in the frequency domain A multi-channel audio encoding method, comprising: a quantization step of performing APCM quantization. 10. The quantization step includes: a first quantization step of APCM-quantizing a frequency-domain multi-channel audio signal using bits allocated in the bit allocation step; and a quantization step in the first quantization step. An inverse quantization step of inversely quantizing the decoded signal to restore the decoded signal; and a stereo for forming a stereo signal by mixing the audio signal restored in the inverse quantization step and the multi-channel audio signal in the frequency domain. A signal forming step; and a second quantization step of further APCM-quantizing a frequency-domain stereo audio signal formed in the stereo signal forming step using the bits allocated in the bit allocation step. The multi-channel audio encoding method according to claim 9.