Smart Transcoding between CELP Speech Codecs through Voiced Oriented Pitch Mapping

Abstract

The deployment of incompatible standardized speech codecs implies interoperability issues between telecommunication networks. Transcoding from one codec format to another is necessary at gateways between networks in order to assure the interoperability. This transcoding reduces speech quality, involves computation load and additive delay. Recently, a fair amount of work has been conducted on studying alternative transcoding methods reducing complexity and delay. In this context this paper further elaborates on solutions based on pitch mapping between standardized CELP codecs. It presents an alternative solution to the already published ones by applying a pitch mapping driven by the voiced property of the frames of analysis. A better compromise between reduction of complexity and impact on the quality of the processed speech signal is thus achieved.