Robust vector quantization in spectral coding
Roar Hagen, P. Hedelin · IEEE International Conference on Acoustics Speech and Signal Processing · 1993
A mapping of a binary block code is used to generate the reconstruction vectors of a vector quantizer. The aim is to secure channel robustness while allowing for efficient design, storage, and handling of the vector quantizer. The general procedure is exemplified by the task of spectral coding for speech transmission. Using an LSP (line spectrum pair)-representation for the spectrum, it is demonstrated that short block codes are feasible, thus allowing for compact storage of the code-book. These short block codes combine low quantization noise in clear channel conditions with robustness to bit errors introduced by a noisy channel. A reduced length code offers the possibility of obtaining a vector quantizer. The block code structure of the code vector generation may also be exploited for efficient search of the code-book.>