This paper presents a strategy for diminishing the correlation of voiced speech using the Waveform Compensation Method (WCM). It introduces a frame matching algorithm that can be used to improve the effectiveness of WCM.


The computational complexity of WCM analysis is significantly lower than the conventional linear prediction method, and the algorithm delay is much shorter. More importantly, WCM is remarkably effective not only at removing short- and long-term correlation of voiced speech, but also at maintaining high speech quality. These advantages make WCM a potentially useful tool in speech coding.