A robust vector quantized sub-band coder for good quality speech coding at 9.6 Kb/s

A vector-quantized subband coder is presented that improves quantization of the subband signals, provides robust noise shaping, and makes a compact real-time implementation possible. Speech is inverse-filtered with a single tap-pitch filter before it is split into sixteen bands. The energy in each band is then computed for use in the noise-shaping process. One sample from each band is used to construct the residual vectors, which are vector-quantized using a single Gaussian codebook. This makes the coder robust to all speakers and allows it to operate in a variable-bit-rate environment