Noise robust pitch extraction based on auto-correlation analysis in the frequency domain
Y. Tabata, Testuya Shimamura · 2002
This paper proposes an improved auto-correlation method for extracting fundamental frequency (pitch) from noisy speech. By transforming the auto-correlation function in the time domain to a synthetic wave consisting of the sum of cosine functions in the frequency domain, we improve the performance of pitch extraction from female speech. For male speech, the performance of the proposed method is competitive with that of the conventional method.