Nonlinear estimation of EGG signals with applications to speech pitch detection

Pitch detection remains a fundamental problem in speech processing. Accurate Pitch information can be obtained with the use of an electroglottograph (EGG), which measures the impedance across the glottis. The differentiated EGG signal clearly marks glottal events. Unfortunately, EGG signals are not available in most practical applications. A novel method of pitch detection is proposed here that requires the EGG signal only during optimization. In operation, the proposed pitch detector marks glottal closures based strictly on the acoustical speech waveform