Efficient pitch filter encoding for variable rate speech processing

Analysis-by-synthesis techniques are used in a wide variety of speech codin g standards and applications for rates below 16 kbps, The presence of a lon g-term predictor, commonly known as the adaptive codebook, is critical to c oder performance at the lower rates. Unfortunately, the encoding rate and c omputational requirements for high-quality encoding of pitch filter paramet ers can be excessive. Several popular approaches explore the trade-off betw een predictor order, allocated bit rate, and computational requirements for long-term predictor optimization. Here, we investigate the relative perfor mance of several longterm predictor structures and present a new approach t o vector quantization of pitch filter coefficients having subjective qualit y equivalent to other schemes, but at a lower coding rate and requiring sig nificantly less closed-loop computation. Performance is evaluated in a vari able-rate CELP coder at an average rate of 2 kbps and in Federal Standard 1 016 CELP.