ANALYZER FOR SPEECH IN NOISE PRONE ENVIRONMENTS

摘要

A voice pattern captured in a noisy environment is analyzed to form a representative sequence of speech parameter signals. The voice pattern affected by noise is divided into intervals by successive time slots. A predictive error signal is produced for each successive interval depending on the utterance affected by noise in the interval and an estimate of the ambient noise. The predictive parameter signals are selected to minimize the predictive error signal of each time slot, so as to produce accurate digital codes representing the utterance.