System and method for sound recognition with feature selection synchronized to voice pitch

摘要

Speech signals are transformed into phoneme identification signals which represent the first two positive slope durations following the maximum peak amplitude of a pitch cycle.