SBIR-STTR Award

A novel signal processing and phonetic analysis system operating in the time domain is proposed which avoids the costly and timeconsuming mathematical software filtering schemes often used. The pitch synchronous ratiometric factors representing phonemic elements of speech are separated from the sporadic noise and certain personality variable features. The purified string of encoded phonetic elements is proposed as input to a voice command system based on a speaker independent lexicon of phonetically organized word and connected phrases which is rather insensitive to speaker stress and background noise.