一种高效的语音分段算法的研究

Study on a high efficiency speech segment algorithm

分段声码器是实现超低速率语音编码的一个重要途径。针对分段声码器的语音分段方法展开了研究,并提出了一种新的分段算法,该算法构造了一个新的度量函数,它在夹角距离的基础上加入了段间距离和增益校正因子,并将参数分级,使其能进行自适应调整以适应噪声环境。仿真结果表明该分段算法精度高,在噪声环境下依然可靠,且延时小,综合性能优于文献[1]提到的分段算法。

Segment vocoder is an important approach to achieve ultra-low-rate speech coding. This paper described the speech segment method of segment vocoder and presented a new segment algorithm, where a new measure function is designed, in which distance between segments and gain correction factor are added to the angle distance and the parameter is scaled to conduct an adaptive adjustment to adapt to noisy environments. The stimulation shows that the algorithm has a high accuracy and is robust in noisy environments and is low-delayed, which can give out evidence of the algorithm ＂s advantage over the segment algorithm presented in the ref.