Enhanced channel coding using source criteria in speech coders

Abstract

Spectral efficiency is deemed highly desirable in many digital voice communication systems. This has largely been achieved by using bit reduced representations of speech signals through advanced speech coding techniques. Transmission errors pose a threat to the quality of voice service in these systems. Forward error correction (FEC) coding techniques is often used to curb the effects of channel impairments, but the performance is limited to bandwidth limitations. This can result in a rapid decline of the speech quality as the effects of the channel impairments increase. We describe a technique for further improving the FEC performance without additional spectral requirements.<>