Fundamental Frequency Extraction of Noisy Speech Using Exponent Enhancement in Weighted Autocorrelation

This research proposes a powerful method for extracting fundamental frequencies from speech in noisy environments that is more successful for speech processing applications. This work discusses a noise-resistant method for fundamental frequency extraction based on an exponent augmentation in the weighted autocorrelation function. To demonstrate the greater accuracy for fundamental frequency extraction, we focus on the exponent of the magnitude difference function. According to experimental results, proposed approach’s presentation in noisy situations provides an unequaled presentation compared to the conventional technique when an appropriate exponent is used.