汉语/英语AMR语音编码参数统计特性

Statistical Characteristics of Chinese and English Speech Coding Parameters for Adaptive Multi-Rate (AMR) Algorithms

摘要: 为研究自适应多速率(AMR)算法应用于汉语语音和英语语音所产生的差异,针对AMR语音编码算法的特点,在分析了大量汉语语音数据的基础上,对算法中的线谱频率(LSF)、基音周期等关键编码参数采用高斯混合模型(GMM)进行了比较并给出相应结论.该研究有助于AMR算法针对汉语语音的性能优化.

Abstract: Difference between the statistical characteristics of adaptive multi rate coding parameters on Hanyu Putonghua (Mandor in) and English is studied. By analyzing a corpus of Hanyu Putonghua speech data, the LSF, pitch and other coding parameters are compared by adopting the Gaussian mixture model (GMM), the corresponding conclusions are drawn. This study can be helpful in the improvement of AMR algorithms for the Chinese language.