NOISY CHANNEL PERFORMANCE OF PREDICTIVE TRELLIS CODED QUANTIZATION

Abstract

This paper investigates the noisy channel performance of a low delay, high-quality speech coder at 16 khits/sec. The system to he studied is based on trellis coded quantization (TCQ). In the work presented, we demonstrate that when trellises based on feedback-free convolutional encoding circuits are employed, error propagation in TCQ systems is not a serious problem. Perfor- mance comparisons are made using a fixed predictor and four different adaptive predictors for both error-free and noisy chan- nels. Informal listening tests reveal that the encoded speech is of good communications even in very noisy channels. In an ideal channel, reconstructed speech of near toll quality can be obtained from a system employing an au-pole least squares lattice predictor with pitch prediction.