Dynamic time-scale compression for waveform speech coding

The bit rates of existing speech coders can be scaled down through the use of time-compression of speech signals. Of course this implies a small increase in coder complexity and some loss of quality. This work focuses on the use of dynamic time-compression in order to reduce the loss of quality. Long, stationary segments do exist in speech, therefore they can tolerate high compression rates without much degradation. Use of the appropriate rate everywhere in time results in increased overall performance from the point of view of both total time-compression factor and resulting signal-to-error ratio.