Switched scalar quantization of a speech signal using forward and backward adaptation

This paper analyzes the switched models of the nonuniform scalar quantization of a speech signal adapted to the average power of a signal. A comparision of the proposed switched quantizers´ performances for the forward and the backward techniques of adaptation in a wide range of signal variance is also presented in the paper. An analysis of the experimental results and the proposed criteria for the models and parameters selection of the quantizer design is conducted with the aim to achieve the highest possible speech signal quality.