Multi-pulse excitation linear predictive coding of speech signals

Abstract

The multi-pulse excitation Linear predictive Coding lMPE-LPC) system produces synthetic speech that is more natural sounding than the classical linear predictive coder. In the the MPE-LPC system, the excitation signal is modeled with a few pulses per frame of speech regardless of whether the frame IS voiced or unvoiced. ~he quality of the synthesized speech improves with the number of pulses used per frame. Pulses are computed by minimizing the weighted square error between the original speech and the synthetic speech. Conceptually, this is a simple problem. However, locating the pulses in the excitation signal can be quite involving when the number of pulses is large. In this study the MPE-LPC system is described and simulated on an IBM PC30XT computer. A computation reduction method in the pulse search process is proposed and compared with other existing techniques. Results show that this method compares well with earlier methods. The effect of the MPE-LPC system parameters on the quality of the reconstructed speech has been investigated and explained. Pulse density reduction by pitch prediction has also been attempted using a one-tap predictor. Results obtained show that pitch prediction improves the quality of the reconstructed speech but not significantly.