论文部分内容阅读
本文介绍了一种编码速率为８ｋｂ／ｓ的高质量实时语音编码器．它采用了代数码本激励线性预测（ＡＣＥＬＰ）的编码方法，并采用高效的码本结构、码本搜索技术和矢量量化技术来获得较高的语音合成质量和较低的算法复杂度．在无需外部ＲＡＭ和ＲＯＭ的情况下，该算法已用ＴＭＳ３２０Ｃ５０实时实现并用于一个实时的全双工通信系统．通过信噪比及人耳主观听觉实验等性能测试表明，该算法的性能明显优于北美的８ｋｂ／ｓＶＳＥＬＰ语音编码算法而算法的复杂度却与ＶＳＥＬＰ相当．
This article describes a high-quality real-time speech encoder with a code rate of 8kb / s. It uses the algebraic codebook excitation linear prediction (ACELP) coding method, and the use of efficient codebook structure, codebook search and vector quantization techniques to obtain higher speech synthesis quality and lower complexity of the algorithm. The algorithm has been implemented in real time with the TMS320C50 and used in a real-time full-duplex communication system without the need for external RAM and ROM. The performance tests of signal to noise ratio and human ear subjective auditory experiments show that the performance of this algorithm is obviously better than that of North American 8kb / sVSELP speech coding algorithm, while the complexity of the algorithm is as VSELP.