JP2002189500A - Audio signal encoding method, audio signal encoder, and recording medium having audio signal encoding program recorded thereon - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simplify calculations performed for bit assignment and to assign bits adaptively also in the case of a low bit rate in an audio signal encoding method. SOLUTION: In an audio signal encoding method, the number of bits assigned to each divided band is determined, and, on the basis of the determined number of bits, an audio signal is quantized and the resultant signal is converted into a coding format determined beforehand. A corrected SNR(signal-to-noise ratio) is determined on the basis of the total number of bits of one frame in the case a number of bits with which an SNR equal to or greater than the SMR(signal- to-mask ratio) in each band is obtained; and the allowed total number of bits of one frame which is determined by the coding format. Bits are assigned to each band on the basis of the corrected SNR.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio signal encoding method, an audio signal encoding device, and a recording medium on which an audio signal encoding program is recorded.

[0002]

2. Description of the Related Art When encoding an audio signal, an audio encoding method for adaptively allocating bits to reduce distortion in a high frequency band of the audio signal and to prevent deterioration of sound quality is disclosed in, for example, Japanese Patent Application Laid-Open No. Hei 8- No. 211899.

[0003] A speech encoding method disclosed in Japanese Patent Application Laid-Open No. H8-212899 discloses a method for encoding one frame (one frame for encoding).
The unit is, for example, ATRAC (Adaptive TRansform
In Acoustic Coding), an input signal of 512 samples) is divided into a plurality of frequency bands having a uniform bandwidth by a plurality of filters and the like, and a noise level and a masking level are determined based on a ratio between a signal level and a masking level in each band. Ratio, and the ratio between the signal level and the minimum audible level, the band having the maximum noise level and the masking level in each band is detected, and the signal level and the noise level are extracted from the band. Bits are allocated based on the ratio, and a ratio between the signal level of the band and the minimum audible level is set as a threshold. For a band having a ratio between the signal level equal to or higher than the threshold and the minimum audible level, a minimum necessary number of bits ( (For example, 2 bits).

[0004]

Problems to be Solved by the Invention
In the speech coding method disclosed in Japanese Patent Application Publication No. 9-209, bits are preferentially allocated from a band having a ratio between a maximum noise level and a masking level in each band, that is, a band requiring a large dynamic range. In addition, a ratio between the signal level of the band and the minimum audible level is set as a threshold, and the minimum necessary number of bits is allocated to a band having a ratio between the signal level and the minimum audible level equal to or higher than the threshold. .

For this reason, when a minimum necessary number of bits is allocated to a band having a ratio between the signal level and the minimum audible level that is equal to or higher than the threshold value of the ratio between the signal level and the minimum audible level, The number of bits required per frame as compared with a case where bits are preferentially allocated from a band having a ratio between a maximum noise level and a masking level (a band requiring a large dynamic range) in the band Increase.

[0006] When the bit rate of the encoding format is high (the permissible number of bits per frame assigned to the audio signal is large), sufficient bits are assigned to each band, and encoding with good sound quality is performed. However, if the bit rate of the encoding format is low (the number of allowable bits per frame allocated to the audio signal is small), bits are not allocated to a band with a small dynamic range such as a signal of a high frequency, and the sound quality is degraded. May be.

The speech coding method disclosed in Japanese Patent Laid-Open Publication No. Hei 8-21899 discloses a speech encoding method for calculating a ratio between a noise level and a masking level in psychoacoustic analysis performed in encoding an audio signal. An operation for calculating the ratio between the signal level and the minimum audible level is required. This complicates the operation for encoding the audio signal. Further, a table for storing the ratio between the signal level obtained by the calculation and the minimum audible level is required.

The present invention simplifies the operation for encoding by not performing the operation using the minimum audible level in the psychoacoustic analysis, and adapts the bit adaptively even when the bit rate of the encoding format is low. An object of the present invention is to provide an audio signal encoding method, an audio signal encoding device, and a recording medium on which an audio signal encoding program is recorded.

[0009]

According to the first aspect of the present invention, there is provided a band dividing step for dividing an audio signal into a plurality of bands, a masking threshold calculating step for calculating a masking threshold for the audio signal, and the masking threshold. A bit allocation step of determining the number of bits to be allocated to each band based on the audio signal, a quantization step of quantizing the audio signal based on the number of bits determined by the bit allocation step, and a step of determining the quantized audio signal in advance. And a format conversion step of converting the signal into a predetermined coding format. The bit allocation step includes the steps of: Allocate the number of bits to obtain the ratio (SNR) A corrected SNR calculating step of obtaining a corrected SNR by correcting the SNR based on the total number of bits of one frame and the allowable total number of bits of one frame determined by the bit rate of the encoding format in the case of Allocating the number of bits to obtain the corrected SNR of the number of bits.

[0010] The invention according to claim 2 of the present application provides a band dividing section for dividing an audio signal into a plurality of bands, a masking threshold calculating section for calculating a masking threshold for the audio signal using auditory characteristics, 1 is determined by the total number of bits in one frame and the bit rate of the encoding format when the number of bits that provides a ratio (SNR) between the signal level and the noise level that is equal to or higher than the ratio of the signal level to the masking threshold (SMR) is assigned. Corrected SN obtained by correcting the SNR based on the allowable total number of bits of the frame
A bit allocating unit that determines R to determine the number of bits to be allocated to each band based on the corrected SNR; a quantizing unit that quantizes the audio signal based on the number of bits determined by the bit allocating unit; And a format converter for converting the audio signal into a predetermined encoding format.

According to a third aspect of the present invention, there is provided a recording medium comprising: a band dividing step of dividing an audio signal into a plurality of bands; and a masking threshold calculating step of calculating a masking threshold for the audio signal using auditory characteristics. , The ratio of the signal level equal to or higher than the ratio (SMR) between the signal level and the masking threshold in each band and the noise level (SMR)
NR) is obtained based on the total number of bits in one frame when the number of bits for obtaining the NR) is allocated and the allowable total number of bits in one frame determined by the bit rate of the encoding format. A bit allocation step of determining the number of bits to be allocated to each band based on the audio signal, a quantization step of quantizing the audio signal based on the number of bits determined by the bit allocation step, and a step of predetermining the quantized audio signal. An audio signal encoding program for executing a format conversion step of converting the audio signal into the obtained encoding format is recorded.

[0012]

FIG. 1 is a schematic diagram showing a schematic configuration of an embodiment of an audio signal encoding apparatus according to the present invention.
In FIG. 1, the audio signal encoding device includes a band division unit 101, a masking threshold calculation unit 102, a bit allocation unit 103, a quantization unit 104, and a format conversion unit 105.

The band dividing section 101 divides an input signal of one frame into predetermined bands. The band division is performed by a filter bank using a plurality of filters or an orthogonal transform such as a discrete cosine transform.

The masking threshold calculator 102 performs frequency analysis on the input signal by means of a fast Fourier transform (FFT) or the like, and calculates a masking threshold using auditory characteristics.

The bit allocating section 103 compares the masking threshold calculated by the masking threshold calculating section 102 with the signal level of each band output from the band dividing section 101 to determine whether the input signal is in a predetermined encoding format. The number of bits to be allocated to each band is determined so that

[0016] The quantizing section 104 determines the number of bits in each band determined by the bit allocating section 103,
Quantize the signal output from 01.

The format conversion section 105 converts the quantized signal into a predetermined encoding format together with bit allocation information obtained from the bit allocation section 104 and outputs the same.

FIG. 2 is a flowchart showing one embodiment of the bit allocation unit in the audio signal encoding method of the present invention. Hereinafter, the ratio between the signal level and the masking level is referred to as the signal masking ratio (SMR).
o), the ratio between the signal level and the noise level is referred to as a signal noise ratio (SNR), and the ratio between the noise level and the masking level is referred to as a noise masking ratio (NMR: No.
ise Masking Ratio).

In FIG. 2, the audio encoding method is as follows.
It has an SMR adjustment value calculation step 201, a corrected SNR calculation step 202, and an allocation bit determination step 203.
The SNR correction step 202 includes an SMR adjustment value calculation step 201 and a corrected SNR calculation step 202.

In FIG. 2, an SMR adjustment value calculation step 201 calculates an SMR in each band on the basis of the masking threshold calculated by the masking threshold calculator 102 and the signal level of each band. Is assigned, the SMR adjustment value C for adjusting the total number of bits of the frame to be encoded (total number of bits) to the allowable total number of bits determined by the bit rate of a predetermined encoding format. Ask for.

Here, the SMR adjustment value calculation step 201
Will be described. FIG. 3 is a flowchart showing the SMR adjustment value calculation step. FIG. 4 is a table showing the relationship between the number of bits used in the bit provisional allocation step and the SNR. FIG. 5 is a schematic diagram for explaining the calculation of the SMR adjustment value C. In FIG. 3, an initial value setting step 3
01 sets an initial value (0 dB) to the SMR adjustment value C. When the bit rate of the encoding format is low,
When a value other than 0 dB (for example, −5 dB) is set as the initial value, the number of calculations for bit allocation is reduced, and the time required for bit allocation processing can be reduced.

The SMR provisional setting step 302 is a step of updating the SMR (SM) by adding the SMR adjustment value C to the SMR for each band.
Rt). When the initial value of the SMR adjustment value C is 0 dB, the SMRt of each band is the ratio (SMR) between the signal level and the masking threshold calculated by the masking threshold calculator 102, as shown in FIG. .

In the bit provisional allocation step 303, provisional allocation (provisional allocation) of the number of bits at which an SNR equal to or higher than the SMR is obtained is performed for each band. Based on the table shown in FIG. 4, as shown in FIG. 5A, as shown in FIG. 5A, by temporarily assigning the number of bits at which an SNR equal to or higher than the SMR is obtained, as shown in FIG. , The number of bits is provisionally allocated.

The total bit number determination step 304 is based on the total bit number of the frame (FIG. 5 (b)) based on the bit number temporarily allocated to each band in the bit temporary allocation step 303.
(The sum of the number of bits in each band shown in (1)) is calculated and compared with the allowable total number of bits.

As a result of the comparison, if the total number of bits temporarily allocated is equal to or smaller than the allowable total number of bits, the setting of the SMR adjustment value C ends. As a result of the comparison, if the total number of bits temporarily allocated is larger than the allowable total number of bits, an SMR adjustment value resetting step 305 is performed.

The SMR adjustment value resetting step 305 is performed so that the total number of bits temporarily allocated approaches the allowable total number of bits and the total number of bits temporarily allocated does not exceed the allowable total number of bits. The SMR adjustment value is decreased by a predetermined amount (for example, 2 dB), and the SMR provisional setting step 3 is performed.
02 is repeated.

In the SMR adjustment value resetting step 305, the SMR adjustment value is reduced by a predetermined amount (for example, 2 dB). In FIG. 5C, by reducing the value of the SMR adjustment value C, the level of the masking threshold increases.

In the SMR provisional setting step 302, the SMR adjustment value C (for example, 2d
SMRt obtained by adding B) is obtained.

The bit provisional allocation step 303 is performed in SMRt
The provisional allocation of the number of bits at which the above SNR is obtained is performed.
Based on the table shown in FIG. 4, by temporarily allocating the number of bits at which the SNR equal to or more than the SMRt shown in FIG. 5C is obtained for each band, the tentative allocation of the number of bits as shown in FIG. An assignment is made.

The total bit number determination step 304 is based on the total bit number of the frame to be encoded (the bit number of each band shown in FIG. 5D) based on the bit number temporarily allocated to each band in the bit temporary allocation step 303. Is calculated and compared with the total number of allowable bits.

As a result of the comparison, if the total number of bits temporarily allocated is larger than the allowable total number of bits, an SMR adjustment value resetting step 305 is further performed, and if the total number of bits temporarily allocated is less than the allowable total number of bits, The setting of the SMR adjustment value C ends.

The processing from the SMR provisional setting step 302 to the SMR adjustment value resetting step 305 is repeated, and in the total bit number determination step 304, when the total number of bits temporarily allocated is equal to or less than the allowable total bit number, FIG. The correction SNR calculation step 202 shown in FIG.

In FIG. 2, the corrected SNR calculation step 2
02 is the SM obtained in the SMR adjustment value calculation step 201.
Based on the R adjustment value C, the initial SNR value (SNR shown in the table of FIG. 4) is corrected, and the corrected SNR (SNR
Find the value of x).

FIG. 6 is a table showing an example of the relationship between the number of bits and SNRx. The SNRx is obtained by adding “−C (SMR adjustment value) × T (constant: 0 <) to the SNR (SNR shown in the table of FIG. 4) used in the SMR adjustment value calculation step 201.
T <1) ”. S shown in FIG.
NRx is a value 10 dB higher than the SNR shown in FIG.

By using the SNR shown in FIG. 4 in the SMR adjustment value calculation step 201 and using the SNRx shown in FIG. 6 in the allocation bit determination step described later,
In some cases, the number of bits due to bit allocation that provides an SNR equal to or greater than the SMR is extremely smaller than the allowable total number of bits determined by the bit rate of the encoding format. Therefore, the constant T is a correction value for bringing the total number of bits of the frame to be encoded closer to the allowable total number of bits determined by the bit rate of the encoding format.

The allocation bit determination step 203 is performed using SNRx shown in the table of FIG.

FIG. 7 is a flowchart showing an allocation bit determination step in the audio signal encoding method according to the present embodiment. 7, the allocated bit determination step includes an NMR calculation step 701, a maximum NMR band extraction step 702, an NMR determination step 703, an NMR update step 704, a total bit number determination step 705, and an update cancel step 706.

The NMR calculation step 701 is based on the SMR of each band, and when the number of bits at which SNRx equal to or higher than the SMR is obtained is assigned,
Rx is subtracted to obtain NMR of each band. At the start of the process, no bits are assigned to any band,
0 dB.

The maximum NMR band extraction step 702 detects a band having the maximum NMR among all the bands.

The NMR decision step 703 is a
It is determined whether or not the NMR of the band has reached the minimum value for allocating the number of bits (the minimum value of the bit allocation, for example, -20 dB lower than the minimum masking threshold). If the NMR of the maximum NMR band has reached the minimum bit allocation, the allocated bit number determination step 203 is terminated. If the NMR of the maximum NMR band has not reached the minimum bit allocation, the NMR is terminated.
An update step 704 is performed.

The NMR update step 704 includes a maximum NMR
The number of bits allocated to the band (bit allocation amount) is increased by one step (one bit), and SM is set based on the increased number of bits.
SNRx is subtracted from R to obtain an updated NMR.

Step 705 for determining the total number of bits
Based on the updated NMR in the update step 704, the total number of bits of the frame to be coded is compared with the allowable total number of bits.
The processing from the maximum NMR band extraction step 702 is repeatedly performed based on the bit allocation amount increased in the update step 704. If the total bit number is larger than the allowable total bit number, the next update cancel step 706 is performed.

The update canceling step 706 is performed by the NMR
The process from the maximum NMR band extraction step 702 is repeatedly executed without updating the bit allocation amount in the update step 705.

By the above-mentioned allocation bit determination processing 203,
The number of bits is assigned to a band having a signal level larger than the masking threshold calculated in the psychoacoustic analysis based on the relationship between the number of bits and SNRx shown in FIG.

FIG. 8 is a schematic diagram for explaining bit assignment. NMR is obtained for all the bands shown in FIG. 8A, and bits are allocated from the maximum NMR band. FIG. 8 (b) shows the result of bit allocation to obtain an SNR equal to or higher than the SMR of each band shown in FIG. 8 (a) based on the table shown in FIG.

When the bit rate of the encoding format is low, the allowable total number of bits allocated to one frame decreases. Therefore, based on the table shown in FIG.
When bit allocation is performed to obtain an SNR equal to or higher than the SMR of each band shown in FIG. 8A, the number of bits allocated to each band is reduced by, for example, 3 bits (bits lower than the dotted line shown in FIG. 8B). The number of bits is reduced, and as shown in FIG. 8C, the number of bits cannot be allocated to a high frequency band.

The high frequency of the audio signal has a relatively low signal level and is easily masked by other signals. When bits are allocated from a band having a large dynamic range, bits are not allocated to a high-frequency signal, and the sound quality is significantly deteriorated.

In order to allocate bits also to a high frequency band, a threshold value of the ratio between the signal level and the minimum audible level is obtained, and a band having a ratio of the signal level higher than the threshold value to the minimum audible level is also determined. There is an audio encoding method for allocating bits.

In this method, the total number of bits in one frame is increased by allocating bits to a high frequency band. Therefore, when the bit rate of the encoding format is low, a low frequency band (dynamic range is low). Less bits are allocated to larger bandwidths)
Also, bits may not be allocated to a high frequency band.

In the present invention, when assigning bits, bits are assigned based on the relationship between the number of bits and the corrected SNR (SNRx) shown in FIG. In the bit allocation based on the relationship between the number of bits and SNRx shown in FIG. 6, the number of bits required to obtain SNRx is
Is smaller than the case shown in FIG.

For this reason, as shown in FIG. 8D, the decrease in the number of bits allocated to each band is reduced.
As shown in (e), the number of bits is adaptively allocated based on the SNRx from a band with a high signal level (a band with a low frequency) to a band with a low signal level (a band with a high frequency). Therefore, even when the bit rate of the encoding format is low, it is possible to prevent the sound quality from deteriorating.

That is, in the above-described audio signal encoding method, when the bit rate of the encoding format is low, the number of bits allocated to the band with a large dynamic range is reduced, and the audio signal is encoded in a band with a small dynamic range (for example, a high frequency). This also increases the chance of assigning the number of bits, reduces the band to which the number of bits is not assigned, and prevents sound quality degradation.

In the above-described audio signal encoding method, since bits are allocated using only the calculated masking threshold, other values (for example, minimum audible threshold) required for bit allocation are set. The calculation process is not required.

The above-described audio signal encoding method calculates an SNRx for various bit rates in consideration of an allowable total number of bits, performs adaptive bit allocation using the SNRx, and converts the audio signal. Can be encoded.

According to the audio encoding method of the present invention, a CD-ROM (Compact Disk Read Only) is used as an audio encoding program executable on a personal computer or the like.
Memory) or the like.

By using the recording medium on which the audio encoding program is recorded, the audio encoding program recorded on the recording medium can be installed in a personal computer, and the audio signal can be encoded using the personal computer. it can.

Therefore, there is no need to use a dedicated audio signal encoding device for encoding the audio signal, and the audio signal can be encoded using a commercially available personal computer.

The audio signal encoding program can be downloaded from a server device via a network and used. A server device connected to the network includes an audio signal encoding program, a terminal device (personal computer) downloads the audio signal encoding program from the server device via the network, and stores the audio signal encoding program in the terminal device. It can be installed and used.

Thus, there is no need to provide a dedicated audio signal encoding device for encoding an audio signal, and the user can go to a store and purchase a recording medium on which the audio signal encoding program is recorded. The audio signal encoding program can be obtained via a network without any hassle.

[0060]

According to the present invention, the operation for encoding is simplified by not performing the operation using the minimum audible level in the psychoacoustic analysis, and even when the bit rate of the encoding format is low. Bit allocation can be performed adaptively.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a schematic configuration of an embodiment of an audio signal encoding device according to the present invention.

FIG. 2 is a flowchart showing one embodiment of a bit allocation unit in the audio signal encoding method of the present invention.

FIG. 3 is a flowchart showing an SMR adjustment value calculation step.

FIG. 4 is a table showing the relationship between the number of bits used in a bit provisional assignment step and SNR.

FIG. 5 is a schematic diagram for explaining calculation of an SMR adjustment value C;

FIG. 6 is a table showing an example of the relationship between the number of bits and a corrected SNR.

FIG. 7 is a flowchart showing an allocation bit determination step in the audio signal encoding method according to the embodiment.

FIG. 8 is a schematic diagram for explaining bit assignment.

[Explanation of symbols]

101: band division unit, 102: masking threshold calculation unit, 103: bit allocation unit, 104: quantization unit, 105: format conversion unit.

Claims (3)

[Claims] 1. A band dividing step for dividing an audio signal into a plurality of bands, a masking threshold calculating step for calculating a masking threshold for the audio signal, and a bit allocation for determining the number of bits to be allocated to each band based on the masking threshold And a quantization step of quantizing the audio signal based on the number of bits determined in the bit allocation step, and a format conversion step of converting the quantized audio signal into a predetermined encoding format. In the audio signal encoding method, the bit allocating step includes a step of determining a ratio (SMR) between a signal level and a masking level in each band.
Based on the total number of bits in one frame when the number of bits giving the above-described ratio of signal level to noise level (SNR) is allocated, and the allowable total number of bits in one frame determined by the bit rate of the encoding format. An audio signal encoding method, comprising: a corrected SNR calculating step of obtaining a corrected SNR in which an SNR is corrected; and an assigned bit number determining step of allocating the number of bits for obtaining a corrected SNR equal to or greater than the SMR for each band. 2. A band dividing unit for dividing an audio signal into a plurality of bands, a masking threshold calculating unit for calculating a masking threshold for the audio signal using auditory characteristics,
The ratio between the signal level in each band and the masking threshold (S
MR) and the ratio (SN) between the signal level and the noise level.
R) is calculated based on the total number of bits in one frame when the number of bits to obtain R) and the allowable total number of bits in one frame determined by the bit rate of the encoding format. A bit allocating unit that determines the number of bits to be allocated to each band based on the audio signal, a quantizing unit that quantizes the audio signal based on the number of bits determined by the bit allocating unit, and a quantized audio signal that is predetermined. An audio signal encoding device, comprising: a format conversion unit that converts the data into an encoded format. 3. A band dividing step of dividing an audio signal into a plurality of bands, a masking threshold calculating step of calculating a masking threshold for an audio signal using auditory characteristics, and a ratio between a signal level and a masking threshold in each band. (SMR) The total number of bits in one frame when the number of bits giving the ratio (SNR) of the signal level to the noise level equal to or higher than the SNR and the allowable total number of bits in one frame determined by the bit rate of the encoding format are A bit allocation step of determining a corrected SNR obtained by correcting the SNR based on the corrected SNR, and quantizing the audio signal based on the bit number determined in the bit allocation step based on the corrected SNR. A quantization step, and converting the quantized audio signal into a predetermined code. Recording medium for recording an audio signal encoding program for executing the format conversion step of converting the reduction format.