JPS6439840A - Coding device - Google Patents

Abstract

PURPOSE:To prevent the deterioration of a tone quality of a reproducing voice by generating the predicting signal of a present time point, based on the difference representative signal and the predicting signal of a previous time point. CONSTITUTION:An adaptive quantization circuit 104 quantizes adaptively an error signal (e) of the present time point by quantization width (w), based on quantization width (w) of the present time point of an output of a quantization width setting circuits 105, and outputs an adaptive PCM coding system (ADPCM) signal (q) of the result. The quantization width setting circuit 105 determines the quantization width (w) of the present time point in accordance with the signal amplitude of the delay ADPCM signal (q) from the previous time point, the quantization width (w) of the previous time point of the previous time point, and a signal S supplied by an adaptive type decoding circuit 108. On the other hand, the adaptive type decoding circuit 108 outputs a decoding representative signal (d) of the ADPCM signal (q) of the previous time point by the quantization width (w), based on the quantization width (w) of the previous time point. Subsequently, a prediction coefficient multiplying circuit 110 multiplies the delayed decoding representative signal (d) by a prediction coefficient (a) (0<a<=1), by which a prediction signal (x) of the present time point is formed. In such a way, the tone quality of a repoducing voice is improved.