Encoding of speech spectral parameters using adaptive quantization methods.