Adaptive Threshold and Modified Adaptive Gain Function based Speech Enhancement Algorithm for Digital Hearing Aid
Deepa Dhanaskodi, C. Poongodi, Shoukath Ali K, T. Perarasi, Thangavel Palaniappan · 2021 Smart Technologies, Communication and Robotics (STCR) · 2021
Speech processing is an important field of signal processing technique in this era due to digitalization of all technologies around us. Processing the real time speech signal needs enhancement because of its acquisition process where the noise will be included in the desired speech signal. If the reference for the noise is available, it will be easily removed from the signal of interest. When the reference signal is not available appropriate speech enhancement algorithm has to be used to clean up undesired components from the signal of interest. The proposed speech enhancement algorithm is focused on an Adaptive threshold and modified adaptive gain function, where the gain function is updated both in frames and frequency bins, this creates the direction to find the speech and noise frames. Modified gain function is processed with the spectrum of input speech spectrum contains noise components, to obtain the enhanced signal. The Speech enhancement method proposed in this paper results with increased Signal to Noise Ratio values and alsoreduced Mean Square Error values for various noisy environment. Time domain plots and spectrum plots are also obtained from the proposed algorithm shows that the algorithm works well on the noisy speech signal to reduce its noise level. Based on the results it is identified that the algorithm proposed in this paper may be used in Digital Hearing Aid to increase the intelligibility of the speech signal and to lessen the listener fatigue condition.