A TIME-VARYING FEATURE SPACE PREPROCESSING PROCEDURE FOR TELEPHONE BASED SPEECH RECOGNITION

摘要

An improved speech recognition system, in which transformation processparameters are generated in response to selected characteristics derived fromspeech inputs obtained from both carbon and linear microphones. Thetransformation process parameters are utilized in conjunction with selecteddigitized speech models to improve the speech recognition process based on thecarbon microphone property of suppressing speech spectral energy for low energyinvoiced sounds, and also for low energy regions of the spectrum between formantpeaks for voices sounds.