摘要
分析了基于CELP语音编码模型的参数转码算法,并以此为基础对关键性技术进行了总结,主要分析和描述了LSP系数、自适应码书、固定码书和码书增益参数的转码算法。设计并实现了一种G.729A8Kb/s到AMR-NB7.95Kb/s模式的参数转码算法,客观质量测试结果表明,该参数转码算法的合成语音质量略优于级联方法。
Several transcoding algorithms are analyzed based on CELP speech coding, and the key techniques are summarized. The transcoding algorithms related to LSP(Line-Spectrum Pair) parameters, adaptive codebook, fixed codebook and their gains are described. A kind of parameter transcoding algorithm between G.729A at 8 Kb/s and AMR-NB at 7.95 Kb/s is designed. The objective quality test results show that this transcoding algorithm performs better than tandem transcoding.
出处
《电声技术》
2009年第4期66-71,共6页
Audio Engineering
关键词
语音编码:语音转码:码激励线性预测
speech coding
speech transcoding
code-excited linear prediction