Fundamental Frequency Extraction by Utilizing the Modified Weighted Autocorrelation Function in Noisy Speech

This research suggests an efficient method that is better suited for speech processing applications for retrieving the accurate fundamental frequencies from speech in noisy conditions. This research describes a modified weighted autocorrelation function for fundamental frequency extraction that is noise-resistant. To gain more accuracy in extracting the fundamental frequency, we have focused on reducing the noise by using the third-order statistics of the noisy speech signal. The proposed fundamental frequency extraction method is composed of two stages: generating third order statistics of speech signal and applying weighted autocorrelation function on it. From experimental results, it is observed that the proposed method shows better performance in noisy environments compared to other conventional methods.