Abstract: Energy-spectrum entropy and subharmonic-harmonic rate methods were used to analyze speech based on speech sinusoidal model. Speech parameters being extracted according to least square method criterion, by introducing multiform quantization method and amplitude parameter fitting quantization model and using dynamic time warping techniques in the process of splitting vector quantization to fast search codebook, a variable-rate low bit rate wideband speech coding algorithm was proposed. Simulation results indicate that the proposed algorithm is characterized by high precision,good real timeness ,and high synthesis speech quality,it also can reconstructed high quality speech in the presence of automobile and music background noises under the condition that speech coding rate and calculation complexity were reduced effectively.

Key words: information processing technology, wideband speech coding, sinusoidal model , spectrum entropy, subharmonicharmonic, LSM, multiform quantization