METHOD AND DEVICE FOR ROBUST PREDICTIVE VECTOR QUANTIZATION OF LINEAR PREDICTION PARAMETERS IN VARIABLE BIT RATE SPEECH CODING

摘要

The present invention relates to a method and device for quantizing linearprediction parameters in variable bit-rate sound signal coding, in which aninput linear prediction parameter vector is received, a sound signal framecorresponding to the input linear prediction parameter vector is classified, aprediction vector is computed, the computed prediction vector is removed fromthe input linear prediction parameter vector to produce a prediction errorvector, and the prediction error vector is quantized. Computation of theprediction vector comprises selecting one of a plurality of prediction schemesin relation to the classification of the sound signal frame, and processingthe prediction error vector through the selected prediction scheme. Thepresent invention further relates to a method and device for dequantizinglinear prediction parameters in variable bit-rate sound signal decoding, inwhich at least one quantization index and information about classification ofa sound signal frame corresponding to the quantization index are received, aprediction error vector is recovered by applying the index to at least onequantization table, a prediction vector is reconstructed, and a linearprediction parameter vector is produced in response to the recoveredprediction error vector and the reconstructed prediction vector.Reconstruction of the prediction vector comprises processing the recoveredprediction error vector through one of a plurality of prediction schemesdepending on the frame classification information.