Um metodo de compressão de audio baseado na codificação de Subbandas Wavelets

AUTOR(ES)
DATA DE PUBLICAÇÃO

2001

RESUMO

This work presents a method for audio compression based on wavelets sub-band quantization and coding, and proposes a coder based on that method. First, the work describes the Wavelet Transform with emphasis on the concepts of multi-resolution analyzes and their relationship with filtering processes and decomposition in sub-bands. A description of the Wavelet Packets Transform is also included since it is used as a decomposition format by the proposed coder. Then, the paper describes the elements of the proposed coder. They include a psychoacoustic model, a bit allocation algorithm, an escalar quantization stage, a vector quantizer, and an entropy coder. The psycho-acoustic model and the bit allocation algorithm are partially based on those used by the MPEG-1 coder in Layers 1 and 2. The proposed coder introduces a method of mapping to place the results of the psychoacoustic model MPEG-1 layer 2 in the domain of the wavelets. In that way, the results of the model will be valid to determine the resolution of the quantizers to be used in coding each sub-band. The proposed coder also introduces a format of vector quantization and an entropy code that is adapted to the behavior of the sub-band samples. Those samples are the audio information to be stored or transmitted. Using these compression techniques the coder presents a good performance for rates around of 128, 96, and 80 kbit/s for mono-channel audio signals. Finally, the paper discusses the results and explains the subjective-type evaluation method used.

ASSUNTO(S)

simulação (computadores digitais) compressão de dados (telecomunicações) matlab (programa de computador) processamento de sinais filtros e filtração sistemas multimidia

ACESSO AO ARTIGO

Documentos Relacionados