Ensuring predictor tracking in ADPCM speech coders under noisy transmission conditions

The problem of predictor mistracking for narrowband signals in adaptive differential pulse-code modulation (ADPCM) speech coders is shown to arise through feedback from the signal reconstruction filter to the predictor adaptation process. A class of residual signal driven lattice predictors is defined, which guarantees tracking for all signals without regard to the order of prediction. Under error-free transmission conditions, a segmental SNR (signal-to-noise ratio) drop for speech of as much as 2 dB may be encountered for this lattice predictor. In more practical telecommunication applications, however, this degradation is outweighed by the enhanced speech and dual tone multifrequency signal transmission performance due to improved robustness in the presence of errors