A greedy adaptive method for time scale modification based on acoustic prediction characteristics of human auditory system

The Synchronized Overlap-Add (SOLA) algorithm neglected the perceptual characteristics of real sound speech signals, when sampling rate was low or scaling proportion was large the process effect would obviously be reduced, aimed at such problems, a greedy adaptive algorithm was proposed through analyzing the acoustic prediction characteristics of human auditory system. This method firstly detected the transition segments of the speech using a subband spectrum entropy measure and applied different scaling factors to the transition segments and non transition segments and put forward an adaptive algorithm. Then it changed the scaling factors locally, the defect of the whole modified proportion was further ameliorated and a greedy adaptive algorithm was created. The simulation showed that the method improved the natural degree of the synthetic speech signals and the scaled time deviation was small. The greedy adaptive algorithm solves the problem of how to improve the quality of the time scaled speech and how to exactly modify the duration of speech signals.