发明公开
EP1132892A1 VOICE ENCODER AND VOICE ENCODING METHOD 有权
Sprachkodierung und -dekodierungssystem

VOICE ENCODER AND VOICE ENCODING METHOD
摘要:
A vector codebook 1094 storing a plurality of typical samples of quantization target vectors is created. Each vector consists of three elements, which are values corresponding to logarithmic values of an AC gain and SC gain and an adjustment coefficient of SC prediction coefficient. Prediction coefficient storage section 1095 stores coefficients to perform predictive coding. These coefficients are MA prediction coefficients and a number of coefficients corresponding to the degree of prediction, of two types, AC and SC, are stored.
Parameter calculation section 1091 calculates parameters necessary for distance calculations from the input perceptual weighted input speech, perceptual weighted LPC synthesis of adaptive code vector, perceptual weighted LPC synthesis of stochastic code vector, further decoded vectors (AC, SC, adjustment coefficient) stored in decoded vector storage section 1096 and prediction coefficients (AC, SC) stored in prediction coefficient storage section 1095.
公开/授权文献
信息查询
0/0