Abstract:
A quantizer and method of an LSF coefficient in a wide-band speech coder using trellis coded quantization algorithm are provided to improve an SD performance and assigned bits by reducing an error transfer as a result of using in parallel both a predictional structure and a non-predictional structure. A quantizer of an LSF coefficient in a wide-band speech coder includes a predictional structure quantizing portion(200), a non-predictional structure quantizing portion(210), and a switching portion(220). The predictional structure quantizing portion calculates a quantized candidate vector by quantizing an LSF coefficient vector, and a predictional quantization final vector of the LSF coefficient vector by trellis coded quantizing the candidate vector with reference to a predicted LSF vector of the LSF coefficient vector. The non-predictional structure quantizing portion calculates a quantized candidate vector by quantizing the LSF coefficient vector, and a non-predictional quantization final vector of the LSF coefficient vector by trellis coded quantizing the candidate vector. The switching portion selects smaller one of differences between the LSF coefficient vector and the predictional and non-predictional quantization final vectors as the final quantization vector of the LSF coefficient vector.
Abstract:
A lossless encoding/decoding apparatus and a method thereof are provided to execute compression on audio signals through bitstreams having small numbers by enhancing the capability of lossless encoding of frequency coefficients. A lossless encoding apparatus includes a bit converter(422), a run length converter(424), and a run length encoder(430). The bit converter generates first bitstreams on respective levels from quantization indexes on frequency coefficients of a current frame. The run length converter generates symbols which are formed by a run length of second bitstreams where the first bit streams are disposed in one row. The run length encoder encodes the symbols to third bitstreams.
Abstract:
1. 청구범위에 기재된 발명이 속한 기술분야 본 발명은 멀티채널 오디오 압축 코덱의 음질 평가 장치 및 그 방법에 관한 것임. 2. 발명이 해결하려고 하는 기술적 과제 본 발명은 멀티채널 압축 코덱의 음질을 평가함에 있어서, 멀티채널 오디오 압축 코덱의 음질에 대한 청취자의 청취 평가 및 통계 처리 과정을 생략하고, 음질에 대한 객관성 및 일관성있는 측정을 통해 멀티채널 오디오 재생 환경에서 선별된 청취자가 느끼는 청각적인 평가를 적절한 방법으로 통계 처리한 것과 유사한 평가 결과를 얻도록 하기 위한, 멀티채널 오디오 압축 코덱의 음질 평가 장치 및 그 방법을 제공하는데 그 목적이 있음. 3. 발명의 해결방법의 요지 본 발명은, 음질 평가 장치에 있어서, 멀티채널 오디오 재생 시스템의 각 채널(L, R, C, LS, RS)로부터 입력된 멀티채널 오디오 신호를 바탕으로 양이 입력 신호를 생성하기 위한 전처리수단; 상기 생성된 양이 입력 신호의 양이 상관 정도 왜곡(IACCDist) 및 출력변수를 산출하기 위한 출력변수 계산수단; 및 상기 산출된 양이 상관 정도 왜곡(IACCDist)과 상기 출력변수를 바탕으로 음질의 등급을 출력하기 위한 인공신경망회로수단을 포함함 . 4. 발명의 중요한 용도 본 발명은 멀티채널 오디오 재생 시스템 등에 이용됨. 객관적 음질 평가, 멀티채널 오디오 압축 코덱, 오디오, 압축, 코덱, 음질 평가, 양이 입력 신호, 양이 상관 정도, 양이 레벨 차이, 양이 상관 정도 왜곡, 양이 레벨 차이 왜곡, 출력 변수, 음질 등급
Abstract:
An apparatus for encoding and decoding by using a converter alternatively according to the correlation of residual coefficients and a method thereof are provided to select a converter with a highest compression rate by perform an RD(Rate-Distortion) cost optimization using DCT(Discrete Cosine Transform) and DST(Discrete Sine Transform) in generating a quantized conversion coefficient through converters and quantizers after prediction between images or within the images in a predetermined size of block, thereby improving the compression rate of an image block. A first converter(31-34) performs the DCT(Discrete Cosine Transform) in a block unit, first quantization, first inverse quantization and IDCT(Interger-approximated Discrete Cosine Inverse Transform) of residual coefficients which are generated after performing prediction between images or within the images. A second converter(35-38) performs the DST(Discrete Sine Transform) in the block unit, second quantization, second inverse quantization, and IDST(Interger-approximated Discrete Sine Inverse Transform) of the residual coefficients. A selection unit(39) performs RD Cost(Rate-Distortion Cost) to select a converter with a high compression rate by block, and a display unit(40) displays converter information selected by the selection unit on a corresponding flag bit in a macro block unit.
Abstract:
An apparatus and a method for restoring a multi-channel audio signal by using an HE-AAC decoder and an MPEG(Moving Picture Experts Group) surround decoder are provided to maintain synchronization between a downmix signal and the MPEG surround decoder by adding a delay unit in connecting the downmix signal outputted from the HE-AAC decoder to the QMF signal of a real number area, the QMF signal of a complex number area, and a time area signal, thereby restoring a desired multi-channel audio signal. An apparatus for restoring a multi-channel audio signal includes a real to complex converter(503) and a delay unit(505). The real to complex converter converts the QMF(Quadrature Mirror Filter) signal of a real number area outputted from an HE-AAC(High Efficiency-Advanced Audio Coding) decoder into the QMF signal of a complex number. The delay unit applies delay generated in the real to complex converter for the QMF signal of the complex number area outputted from the HE-AAC decoder.
Abstract:
A system and a method for encoding and decoding multi-channel audio are provided to maintain compatibility with an existing multi-channel audio receiver and encode and decode a multi-channel audio signal of high quality by using SAC(Spatial Audio Coding) and SBR(Spectral Band Replication). A bit stream demultiplexing unit(301) demultiplexes a multiplexed bit stream into a multi-channel audio bit stream of a low sampling frequency, an SAC bit stream, and an SBR bit stream. A multi-channel audio decoding unit(303) decodes the multi-channel audio bit stream and outputs a down-mix signal and a multi-channel audio signal. An SBR decoding unit(305) decodes the down-mix signal and the SBR bit stream by an SBR type, and outputs a down-mix signal whose high frequency region is recovered. An SAC decoding unit(307) decodes the multi-channel audio signal by an SAC type by using a spatial queue included in the SAC bit stream and the down-mix signal whose high frequency region is recovered, and outputs a multi-channel audio signal of a high sampling frequency.