Adaptive Harmonic Plus Noise Model (A-HNM) Analysis for Noise Reduction in Speech Signal

Speech signal enhancement has become a popular study area during the past few years. A plethora of techniques have been developed to improve the perceptibility and quality of speech signals. These techniques are frequently designed to restore the original signal from noise-damaged data. The wavelet transforms are commonly employed for processing speech signals. This research work proposes an Adaptive Harmonic plus Noise Model (A-HNM) analysis for removing the noise in speech signals. Initially, the speech signals from standard datasets are gathered and further, decomposed with the Empirical Mean Curve Decomposition (EMCD) technique. Then, the noise reduction is made over the decomposed signals with the help of A-HNM analysis, where the HNM parameters are tuned with the Red Deer Algorithm (RDA) to reduce the noise in the damaged speech signal and improve the performance. Additionally, the post-processing using wiener filtering is performed to increase the efficacy of standard improvement approaches.