On the use of wavelets and cepstrum excitation for Pitch Determination in real-time

Abstract

In the current paper, we propose a new pitch tracking technique based on a wavelet transform in the temporal domain. Our algorithm is designed to determine the pitch frequency of the speech signal using a simple voicing decision algorithm. The pitch period is extracted from the cepstrum excitation signal processed by a wavelet transform; then the pitch contour is refined by thresholding and correction algorithms without any postprocessing. The results obtained show that the proposed algorithm provides very good pitch contours compared to those furnished by the Bagshaw database.

Improved Algorithm for Pitch Detection and Harmonic Separation

“…we have got in time domain is T0, pass through this frame with a time-varying filter as is defined in (7). The length of the transformation is T0.…”

“…The short-time Fourier transformation for the ith point of a frame can be estimated as: Where g (n) varies with different pitch periods of different frames. Then we design the time-variant filter according to the spectrum we get from (7) to track the path of the fundamental frequency. We can get time domain signal s'(n) through making inverse discrete Fourier transformation (IDWT) of residual spectrum.…”

“…T-F transformation method means to transform the speech signal of time domain to frequency domain or spectral domain in order to estimate its pitch period. This method is more complex but has a better effect than others [7]. Inspired by normalized autocorrelation function [8] and pitch detection algorithm in MBE vocoder [9], we have proposed an improved algorithm of pitch detection in this paper.…”