A Novel Speech Enhancement Solution Using Hybrid Wavelet Transformation Least Means Square Method
Jagadish S. Jakati, Shridhar S. Kuntoji · International Journal of Engineering Trends and Technology · 2021
Currently, minimizing the noise in speech or audio signals is a challenging issue in the field of speech recognition, speech enhancement, and other speech communication applications. These applications have fascinated research community due to their diverse use in real-time, online and offline applications. Several approaches have been presented to enhance the quality of speech. Currently, the Wavelet Transformation based approach and Least Means Square based filtering schemes are extensively adopted in various researches. The existing techniques suffer from computational complexity and performance related issues. Thus, we focused on combining these schemes and presented a hybrid approach that uses wavelet packet transform and an adaptive LMS scheme. We present an extensive simulation study and comparative analysis by using the NOIZEUS speech corpus database. The experimental analysis shows a substantialaugmentation in the performance of speech enhancement.