Pitch detection based on signal-to-noise-ratio estimation and compensation for continuous speech signal

Abstract

On speech signal processing, it is very important to find the pitch period of voice. The reason is why it is used in variable places, such as speech-enhancement-system, speech-recognition system, speaker-classification- system, and handicapped assisting-system. It is, however, difficult to detect the accurate pitch on the low SNR section caused by the voice transition or the corrupted signal by noise. In this paper, we propose the method which is able to detect the more accurate pitch based on the Signal to Noise Ratio(SNR) estimation by the SNR compensation for continuous speech signals. We show the improved performance of the proposed pitch detection by the implemented results for the pitch contour of variable SNR signals. © 2012 Springer-Verlag.