TH28599B - Prefixes and methods for coding speech - Google Patents

Abstract

Prefixes and methods for coding speech In which The bitrate of the encrypted data can be Get variability The output vector x will be quantized. The vector titration is weighted by the 502 vector quantization unit of the first 500 quantization unit of the 500 vector quantization unit. The quantization error X0 'will be subtracted from the source vector x at plus 505.The vector of the quantization error y will be dimensionally divided by the unit of quon. Times vector 510 second unit Contour index data at the cliff To quantify the weighted vector which As a result, it is outputted at the quantized 512 1, 512 output connector.Y1 'and Y 2' are mutually assembled and combined by a plus 513. Corresponding to the value Quantized X0 'The quantized value X1 is output.

Claims (1)

1. Speech coding method for dividing the input speech signal on the timeline into blocks. So it is this unit and the encoding of the signal obtained Which consists of steps of Finding the short-term prediction reside, at least for The audio portion of the input audio signal; Also for sine analysis type coding parameters. Based on short-term prediction errors that have been sought like this one; and an absolutely weighted vector quantization operation Relative to sine-shaped analysis type coding parameters; And the non-coding: