摘要
共振峰声码器因其在理论上具有最低码率而一直是参数语音编码算法研究的重点。共振峰编码器的关键算法是基频和共振峰等语音参数的提取。在高分辨率语谱图基础上,利用语音信号的频域特性设计了一种简单有效的基频和共振峰提取算法。通过评价重建语音信号的音质,证明了参数提取算法的准确性。根据语音实验确定编码参数包含基频和前4个共振峰,并在保证语音质量的前提下制定各参数的量化指标。应用实际语音信号对算法的性能进行测试,试验结果证明算法在码率为1 400 b/s时具有良好的语音质量。
Formant vocoder, which has the least coding rate on the theory,is the focus of speech coding. The fundament frequency and formant extraction is the first and the most important step in formant vocoder. An algorithm for the fundamental and formants extraction is designed in accordance with the future of speech spectrum. The algorithm is tested by reconstructing the speech signal using the value of fundamental frequency and formants which is extracted,and the result shows that the parameters＇alue is accuracy. The fundamental frequency and four formants is coded, the quantification of each parameters is made after speech experimentation. The vocoder has good quality when the coding rate is 1 400 b/s.
出处
《现代电子技术》
2007年第21期168-170,共3页
Modern Electronics Technique
基金
江苏省高校自然科学基础研究项目(06kjb510089)
关键词
共振峰语音编码
基频
共振峰
谐波合成
formant vocoder
fundamental frequency
formant
harmonics model
