This project develops and evaluates new speech coding algorithms for voice communication at rates between 2.4 and 8.0 kilobits per second. Particular emphasis is given to algorithms that are robust to transmission bit errors caused by noisy channels such as, for example, mobile radio communication systems. The algorithms under study use basic speech compression techniques such as vector quantization, analysis.by.synthesis, perceptual weighting, short. and long.term linear prediction, and vector excitation coding as well as several newer techniques. This work is important to communication technology particularly in mobile radio systems where bandwidth is a precious resource and relatively high error rates must be tolerated due to limited radio frequency power constraints.