Parameters quantization in sinusoidal speech coder on basis of human auditory model

In the given paper a method of the parameter quantization in speech coder on basis of human auditory model is presented. Output speech parameters of the coder are three sinusoidal parameters: amplitude, frequency and phase. For amplitude and frequency quantization we propose to use a vector quantization, and for phases – scalar quantization. This parameter quantization method allows to achieve bit rate from 3 to 8 kbps depending on the reconstructed speech quality.