Improved HCDM systems with variable rate sampling and silence detection

Abstract

In this thesis work, two improved hybrid companding delta modulation (HCDM) systems are studied. In the first half, the HCDM with variable sampling rate is studied. The theoretical analysis of HCDM with variable sampling rate is done. The methods for sampling rate conversion and efficient buffer control are studied. To optimize the system parameters, computer simulation has been done with various real speech. It has been found that the performance of HCDM with variable sampling rate is about 3 to 4 DB better than that of a conventional HCDM. In addition, design of a hardware system has been considered for the HCDM with variable sampling rate. In the second half, the HCDM with silence detection is studied. For silence/speech detection, a simplified method of Un and Lee``s speech detection algorithm has been used. This algorithm is based on the LDM bit alternation rate. Computer simulation shows that a large SQNR gain of 5 DB can be obtained by coding speech with silence detection incorporated. Design of hardware system for HCDM with silence detection has also been studied.