Improvement and optimization of MELP speech coder by digital signal processor
MELP低速率语音编码器在DSP上的改进与优化*

In the mixed excitation linear prediction(MELP) algorithm,frames include both the voiced and unvoiced sound result in parameter estimation errors and quality degradation for synthesized speech,which is a serious problem.In order to solve it,this paper gave a new approach to analyze and synthesize the transition segments which were called the transition frames model(M-MELP).It could improve the accuracy of parameter estimation and guarantee the quality for synthesized speech.Moreover,it proposed many optimiz...