SPEECH DECODER

摘要

Short-term prediction code is produced through an analysis for a speech signal having a predetermined frame length and extraction of a spectral parameter of the speech signal. In an adaptive codebook each past excitation signal of a maximum integer lag length is stored. The excitation signals read out from the adaptive codebook is delayed in a predetermined pitch frequency range. Calculation of a delay value minimizing the weighted square error as a measure of the evaluation and gains corresponding to the delayed past excitations and searching an adaptive codevector as the optimum lag code representing the pitch correlation of the speech signals is performed. In an excitation codebook excitation codevectors as quantization codes representing residue signals after the long-term prediction is stored. The optimum excitation codevector is determined from the excitation codebook. The long-term prediction including lag code decimating and varying operation for varying the lag codes through decimating thereof.