SIGNAL PROCESSING FOR IMPROVING SPEECH IMIRMENTS

THE LONG-TERN GOAL OF PHASE I AND PHASE II RESEARCH IS TO PROVIDE A NEW TOOL FOR SPEECH THERAPY BASED ON THE DISPLAY OF VOCAL TRACT CROSS-SECTIONS. THE SIGNAL PROCESSING APPROACHES UNDER INVESTIGATION ARE BASED ON SPEECH RECOGNITION TECHNIQUES. THIS SYSTEM WOULD PERMIT USERS TO SEE IN A CONCRETE, GRAPHIC WAY HOW CLOSELY THEY APPROACH A TARGET SPEECH PATTERN AND WOULD PROVIDE SOME SUGGESTIONS ABOUT HOW TO IMPROVE SPEECH PRODUCTION. THE TECHNOLOGICAL INNOVATIONS INHERENT IN THE PROPOSED DESIGN INCLUDE (1) NOVEL APPROACHES TO THE DISPLAY OF ACOUSTIC INFORMATION; (2) NEW ALGORITHMS FOR ACOUSTIC SIGNAL PROCESSING BASED ON PERCEPTUAL CONSIDERATIONS; AND (3) IMPLEMENTATION OF THE SOFTWARE ON A MACINTOSH COMPUTER WITH ITS EXCELLENT GRAPHIC CAPABILITY. PHASE I WILL FOCUS ON THE APPLICATION OF THE SOFTWARE FOR THE PRODUCTION AND STUDY OF CONTINUANTS. IT WILL INCLUDE ACTUAL TRIALS AT THE SPEECH, HEARING AND NEUROSENSORY CENTER OF CHILDREN'S HOSPITAL IN SAN DIEGO TO DETERMINE THERAPIST AND PATIENT ACCEPTANCE OF THE SYSTEM IN AN ITERATIVE PROCESS OF USER-MODIFIED DESIGN.