KR970031362A

KR970031362A - Digital audio coding method

Info

Publication number: KR970031362A
Application number: KR1019950039874A
Authority: KR
Inventors: 김연배
Original assignee: 김광호; 삼성전자 주식회사
Priority date: 1995-11-06
Filing date: 1995-11-06
Publication date: 1997-06-26

Abstract

본 발명은 디지탈 오디오 부호화방법에 관한 것으로서, 입력되는 오디오 데이타를 n개의 주파수대역으로 분류하는 매핑과정; 마스킹 현상과 임계대역들을 고려하여 마스크된 문턱치를 구하고, 각 주파수대역의 신호의 크기와 마스크된 문턱치와의 비를 이용하여 중요도에 따라서 각 주파수대역의 신호에 대하여 비트를 할당하는 비트할당과정; 복원시 발생되는 오차가 최소가 되도록 각 주파수대역의 신호의 특성에 따라서 비트할당된 각 주파수대역의 신호를 소정의 양자화단계로 양자화하는 양자화과정; 및 양자화된 신호를 비트스트림으로 형성하는 비트 패킹과정으로 구성된다. 따라서, 청각심리모델을 사용하는 종래기술이 복원 음질은 좋으나 처리비용 및 처리시간이 많이 소요되는 것에 비하여 청각심리모델을 적용하지 않고도 디지탈 오디오기기의 비트할당에 인간의 음향심리를 고려해 주며, Sm4R을 직접 계산하지 않으므로 하드웨어의 가격과 복잡도를 줄일 수 있을 뿐아니라 데이타의 압축연산시에 필요한 계산량을 줄임으로써 처리속도를 높일 수 있으므로 실시간 처리가 가능하다.The present invention relates to a digital audio encoding method, comprising: a mapping process of classifying input audio data into n frequency bands; A bit allocation process for obtaining a masked threshold in consideration of masking and critical bands, and allocating bits for signals of each frequency band according to importance using a ratio between a signal size of each frequency band and a masked threshold value; A quantization process of quantizing the signals of each frequency band to which bits are allocated in a predetermined quantization step according to the characteristics of the signals of the respective frequency bands so that an error generated during restoration is minimized; And a bit packing process of forming the quantized signal into a bitstream. Therefore, compared to the conventional art using the psychoacoustic model, the sound quality of the reconstruction is good, but the processing cost and the processing time are high. Since it does not calculate directly, it not only reduces the cost and complexity of hardware, but also increases the processing speed by reducing the amount of computation required when compressing data, thereby enabling real-time processing.

Description

Digital audio coding method

본 내용은 요부공개 건이므로 전문내용을 수록하지 않았음Since this is an open matter, no full text was included.

제2도는 마스컴 현상을 설명하기 위한 도면.2 is a diagram for explaining the mask phenomenon.

제3도는 본 발명에 의한 디지탈 오디오 부호화방법이 적용되는 디지탈 오디오 부호화기의 구성도.3 is a block diagram of a digital audio encoder to which the digital audio encoding method according to the present invention is applied.

Claims

A mapping process of classifying the input audio data into n frequency bands; Bit allocation that obtains a masked threshold value in consideration of masking phenomenon and critical bands, and allocates a bit to a signal of each frequency band according to importance using a ratio between a signal size of each frequency band and a ratio of the masked threshold value process; A quantization process of quantizing the signals of each bit-allocated frequency band in a predetermined quantization step according to the characteristics of the signals of the respective frequency bands so that an error generated during restoration is minimized; And a bit packing process of forming the quantized signal into a bit stream.

2. The method of claim 1, wherein the size of the signal in the bit allocation process is divided into frequency bands by the mapping process, and further comprising: a first step having a largest value among signals existing in each frequency band; A second step of converting the audio data into a frequency band by conversion and finding a largest value in a frequency domain corresponding to a processing band of a time domain; And a third step of setting a larger value of the two values found by the first step and the second step to the magnitude of the signal.

3. The method of claim 2, further comprising: obtaining a magnitude of each signal for each frequency band by converting the audio data in the bit allocation process in the bit allocation process; Obtaining a tone component using a magnitude of a signal of an adjacent band of the frequency bands; A third step of obtaining a noise component for each critical band using the remaining components except the tone component; A fourth step of calculating a degree of masking for each band by the tone component obtained in the second step; A fifth step of calculating a degree of masking for each band by the noise component obtained in the third step; And a sixth step of setting the largest value as the final masked threshold of each band by comparing the degree of masking calculated by the fourth and fifth steps with the absolute threshold of each frequency band. Digital audio coding method characterized by the above-mentioned.

4. The method of claim 3, wherein in the second step, the tone component comprises: a first step of setting the signal to a local maximum when the signal in the respective frequency bands is larger than a signal in which the signal is left and right; And a second step of setting this signal as a tone component if a value greater than a predetermined level is greater than another local maximum value within a predetermined range in consideration of a critical band among the local maximum values obtained in the first step. Digital audio coding method.

2. The method of claim 1, further comprising: dividing the magnitude of the signal into frequency bands by the mapping process and finding a value having the largest absolute value among the signals existing in each frequency band; And a second step of determining the magnitude of the signal of each frequency band using a scale factor table as a reference for normalization.

The digital audio encoding method of claim 1, wherein the bit allocation process allocates bits using a signal size in each frequency band and a signal-to-noise ratio for the allocated quantization step.

※ Note: The disclosure is based on the initial application.