MULTI-STAGE VECTOR QUANTIZATION APPARATUS AND METHOD FOR SPEECH ENCODING

摘要

A CELP speech coder/decoder capable of accepting a quantization algorithm in a quantizing section using a multi-stage predictive vector quantization method. The vector quantization of a target is carried out in the multistage predictive vector quantization. The target is obtained and is further vector-quantized. A vector of a first target vector is predicted by a predicting section. The searching section carries out the decoding of the vector using decoded vectors from vector codebooks based on the codes for codevector A and codevector B. The amplitude is obtained from the amplifer storing section and past-decoded vectors stored in the state storing section. The apparatus and method reduce the coded distortion. The apparatus may be employed in speech coding and/or decoding for music, images, or in cellular phones.