A SELP Algorithm Aimed at Real-Time Implementation (NOT PUBLISHED)

A Stochastically Excited Linear Predictive Voice Coder[1] with application in the 4.8 to 9.6 Kb/s range is described. The emphasis of this work is on limiting the computational complexity and memory requirements, with a minimum reduction of speech quality. Currently, work is in progress to implement this coder on real-time floating point hardware. Results from formal listening tests are provided. Tapes will be played during the presentation to demonstrate the speech quality of our software implementation.