论文部分内容阅读
混合激励线性预测MELP编码方法结合混合激励、多带激励以及原型波形内插等编码方法的优点,成为低码率语音编码的研究热点,该方法可以很好的模拟自然语言,在低速率上合成出较高质量的语音,本文以该算法作为基本模型,对参数的内插、高效的矢量量化、多帧联合几项关键技术进行了详细的分析,给出了实现1.2kbps语音编码的算法,并对算法进行了仿真,从仿真结果可以得出该算法具有可行性。
Hybrid excitation linear prediction MELP coding method combines the merits of hybrid excitation, multi-band excitation and prototype waveform interpolation, and becomes a research hotspot in low bit rate speech coding. This method can well simulate natural language and synthesize at low rate This paper takes the algorithm as a basic model, analyzes several key technologies such as parameter interpolation, efficient vector quantization and multi-frame combination in detail. An algorithm of 1.2kbps speech coding is given, The algorithm is simulated, and the simulation results show that the algorithm is feasible.