JP2000293200A - Audio compression coding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the amount of computations by eliminating the computational processes required for quantization/coding if the same state continues in terms of a frame unit in audio data. SOLUTION: One frame of audio data Xn(i) (i=1 to 1,152) is taken in and audio data Xn(i) of less than a threshold value are set to 0. Then, audio data of a previous frame and a present frame are compared (a step S14). If they agree with each other, audio bit stream of the previous frame are replaced and outputted (a step S16), and the processes of an audio model weighting section and a quantization/coding section are omitted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an MPEG (Moving Picture Image Coding Experts
up), and a voice compression encoding method.

[0002]

2. Description of the Related Art In recent years, high-efficiency encoding of moving images and audio (audio) for the purpose of application to large-capacity media has been attempted. MPE is an international standard for this high efficiency coding.
G, which is described in detail in ISO / IEC11172.

As shown in FIG. 2, an audio compression encoding apparatus for encoding audio digital data by MPEG includes a subband analysis section 21, a quantization / encoding section 22, a bit multiplexing section 23, and an auditory model weighting section 24. It is comprised including. The sub-band analyzer 21 converts the audio data of each frame into 32 sub-band samples. On the other hand, the auditory model weighting unit 24 determines bit allocation (bit allocation) for each subband and supplies the bit allocation (quantization / coding unit 22). Based on the result, the quantization / encoding unit 22 quantizes and encodes the subband samples output from the subband analysis unit 21. Then, the bit multiplexing unit 23 outputs the frame-packed high-efficiency coded data as an audio bit stream, and stores it on a recording medium or transmits it to a decoding device on the receiving side.

A model 1 and a model 2 exist as signal processing methods in the auditory model weighting unit 24. The signal processing method of the model 1 is as shown in FIG. That is, in the first step S1, the FFT is performed on the input audio signal.
Make an analysis. Then, in step S2, the sound is divided into a plurality of sub-bands, and the sound pressure of each sub-band is calculated. In the next step S3, the data of each sub-band is selected into a pure tone component and a non-pure tone component, and in step S4, a low-level frequency component,
For example, thinning out data centering on non-pure tone components is performed.
In step S5, an individual masking threshold is calculated, and in step S6, an overall masking threshold as viewed from all subordinates is calculated. Then, proceeding to step S7, the minimum masking level of each sub-band is determined, and the signal-to-mask ratio (SMR) is calculated in step S8. As described above, the SMR is obtained by the auditory model weighting unit 24, and the bit allocation is determined based on the obtained SMR by the quantization / encoding unit 2.
2 is instructed.

[0005]

As described above, speech compression encoding is performed on a frame basis, but the amount of calculation is enormous. In particular, the amount of calculation in the auditory model weighting unit 24 is large. If the audio data of the frame currently being processed is the same as the audio data of the previous frame (the same audio waveform),
The output of the bit multiplexing unit 23 should be the same audio bit stream. Even in such a case, the conventional audio compression encoding method has a problem that a series of encoding processes must be performed and the same operation must be repeated again.

SUMMARY OF THE INVENTION The present invention has been made in view of such a conventional problem, and when the audio data of the current frame is considered to be the same as the audio data of the previous frame,
It is an object of the present invention to realize a voice compression encoding method capable of greatly reducing the amount of computation in the encoding process by omitting a series of encoding processes and using the audio bit stream of the previous frame as it is.

[0007]

According to the first aspect of the present invention, an audio signal is sampled at a specified frequency,
The obtained audio data is framed for each predetermined number of samplings, subband analysis is performed for each frame, audio data of each subband is weighted based on an auditory model, and quantization and quantization are performed based on the obtained weighting. An audio compression encoding method for encoding to generate an audio bit stream, comprising: storing, in a memory, audio data of a previous frame and an audio bit stream in which the audio data of the previous frame is quantized and encoded; When the audio data of the current frame is input, the audio data of the current frame is compared with the audio data of the previous frame held in the memory, and in the comparison, the audio data of the current frame and the audio data of the previous frame have the same audio waveform. Read the audio bit stream of the previous frame from the memory It is characterized in that the output as audio bitstream of the current frame.

According to such a method, when the audio waveform of the previous frame is the same as the audio waveform of the current frame, the arithmetic processing in the encoding process can be greatly reduced.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A speech compression encoding method according to an embodiment of the present invention will be described below with reference to the drawings. The configuration of the audio compression encoding apparatus is the same as that shown in FIG. 2, but the signal processing method is different from the conventional example. FIG.
4 is a flowchart showing a speech compression encoding method according to the present embodiment. Hereinafter, the description will be limited to the encoding process called layer 2 of MPEG1 audio.

In FIG. 1, in step S11, audio data for one frame sample is fetched, and this data is defined as X _n (i) (i = 1, 2,..., 1152, n = frame number). 32 kHz, 44.1 kHz, 48 kHz as sampling frequencies for capturing audio data
There is z. In the next step S12, it is checked whether or not all the audio data X _n (i) of one frame is equal to or less than a predetermined threshold. If it is equal to or less than the threshold value, the current frame is regarded as a noise component only, and the value of the audio data X _n (i) is set to 0.

In the next step S13, it is checked whether or not the previous frame does not include an extra slot. When the sampling frequency is 32 kHz or 48 kHz, no extra slot occurs due to the relationship between the number of samplings 1152 and the number of bytes allocated to one frame. However, when the sampling frequency is 44.1 kHz, the number of bits of a stream of one frame is not constant (fraction occurs), and an extra slot (1 byte) may be included. When the previous frame includes an extra slot as described above, the stream of the current frame cannot be replaced with the stream of the previous frame.
Then, the process branches to 5 to perform a normal encoding process. Then, the process proceeds to step S19 to create data of a new stream.

On the other hand, if it is determined in step S13 that the previous frame does not include an extra slot,
Branch to 14. In this step S14, the audio data X _n (i) of the current frame and the audio data X
_It is checked whether or not _n-1 (i) (i = 1 to 1152) matches. Therefore, the audio data X _{n-1 of the} previous frame is
(I) is stored in a memory, and X _n (i) and X _n−1
(I) is compared.

Here, assuming that the difference between X _n (i) and X _n-1 (i) is Y (i), if the value of Y (i) becomes 0 regardless of the sampling point i, , voice data of the previous frame X _n-1 (i) and the current frame speech data X _n
(I) are considered the same. If the audio data of the previous frame and the audio data of the current frame do not match in this step S14, the flow branches to the above-described step S15, where normal encoding processing is performed. If the audio data of the previous frame and the audio data of the current frame match in step S14, the flow branches to step S16, where the bit stream for the current frame is replaced with the bit stream of the previous frame stored in the memory in advance.

In the next step S17, it is checked whether or not the current frame includes an extra slot. If the previous frame does not include an extra slot and the current frame is a frame including an extra slot, the process proceeds to step S18, where 0x00 is added to the end of the stream of the current frame, and the process ends. If the current frame does not include an extra slot in step S17, the process ends.

In the case where the silent state continues in the frame unit or the same pure tone continues, the sound data of the previous frame and the current frame are regarded as the same. Becomes unnecessary, and the processing of the quantization / encoding unit 22 becomes unnecessary.

[0016]

As described above, according to the audio compression encoding method of the present invention, the audio data of the previous frame and the audio data of the current frame are compared. If the audio data of the current frame matches the audio data of the previous frame, the previous stream can be replaced without performing a series of encoding processes, so that the amount of calculation can be greatly reduced.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a signal processing procedure of a voice compression encoding method according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a voice compression encoding device.

FIG. 3 is an explanatory diagram of a signal processing method of an auditory model weighting unit in the audio compression encoding device (MPEG audio encoder).

[Explanation of symbols]

 Reference Signs List 21 Subband analysis unit 22 Quantization / encoding unit 23 Bit multiplexing unit 24 Auditory model weighting unit

Claims (1)

[Claims] An audio signal is sampled at a specified frequency, obtained audio data is framed for each predetermined number of samplings, subband analysis is performed for each frame, and audio data of each subband is based on an auditory model. A voice compression encoding method for generating an audio bit stream by performing quantization and encoding based on the obtained weights, wherein the audio data of the previous frame and the audio data of the previous frame are quantized. And when the encoded audio bit stream is stored in the memory, and when the audio data of the current frame is input, the audio data of the current frame is compared with the audio data of the previous frame held in the memory. When it is determined that the audio data of the frame and the previous frame have the same audio waveform, An audio compression encoding method, comprising reading an audio bit stream of a frame from the memory and outputting the audio bit stream of the current frame.