Real-time pitch detection on the PDA for cochlear implant applications

Abstract

There is currently a need in cochlear implants to develop speech coding algorithms that provide better access to pitch cues, known to be critical for music perception. Such algorithms require an estimate of the fundamental frequency (FO). This paper presents the implementation of a real-time pitch (FO) detector on a personal digital assistant (PDA). The pitch detection algorithm is based on the autocorrelation function and is implemented real-time on a Dell AXIM pocket PC. Its performance, in terms of FO accuracy, is compared against that obtained by the pitch detection algorithm used in STRAIGHT. The implementation details and real-time performance measurements are also provided.