Open AccessFrequency Controlled Noise Cancellation for Audio and Hearing Purposes
Author(s) -
Ali O. Abid Noor
Publication year - 2020
Publication title -
emitter international journal of engineering technology
Language(s) - English
Resource type - Journals
eISSN - 2443-1168
pISSN - 2355-391X
DOI - 10.24003/emitter.v8i1.473
Subject(s) - noise (video) , computer science , hearing aid , distortion (music) , electronic engineering , speech recognition , telecommunications , acoustics , physics , bandwidth (computing) , engineering , artificial intelligence , amplifier , image (mathematics)
Methods for hearing aids sought to compensate for loss in hearing by amplifying signals of interest in the audio band. In real-world, audio signals are prone to outdoor noise which can be destructive for hearing aid. Eliminating interfering noise at high speed and low power consumption became a target for recent researches. Modern hearing compensation technologies use digital signal processing which requires minimum implementation costs to reduce power consumption, as well as avoiding delay in real time processing. In this paper, frequency controlled noise cancellation (FCNC) strategy for hearing aid and audio communication is developed with low complexity and least time delay. The contribution of the current work is made by offering a method that is capable of removing inherent distortion due filter-bank insertion and assigning adaptive filtering to a particular sub-band to remove external noise. The performance of the proposed FCNC was examined under frequency-limited noise, which corrupts particular parts of the audio spectrum. Results showed that the FCNC renders noise-immune audio signals with minimal number of computations and least delay. Mean square error (MSE) plots of the proposed FCNC method reached below -30 dB compared to -25 dB using conventional sub-band method and to -10 dB using standard full-band noise canceller. The proposed FCNC approach gave the lowest number of computations compared to other methods with a total of 346 computations per sample compared to 860 and 512 by conventional sub-band and full-band methods respectively. The time delay using FCNC is the least compared to the other methods.