Multi stage filter bank system for improved noise reduction in hearing aids
M. Poornima, E. Rajinikanth · 2016
Signal processing a major tool used for noise reduction. Here we discussed noise reduction, frequency, and amplitude filters for a digital hearing aid (DHA) which can be configured as per the needs. Our digital hearing aid overcomes the disadvantages of the traditional analog model. We simulated our filters and successfully reduced white noise, increased the gain for frequencies which were unable to hear and shaped the amplitude to prevent any of the frequencies from becoming too loud. The filter-bank system available in hearing aids has to satisfy various constraints such as and high stop-band attenuation and low latency, usually for low frequency resolution. Improper frequency resolution leads to residual noise artifacts since in speech signals the spectral harmonics is not properly handled in frequency-domain noise-cancellation algorithms. Especially when considering the female speech signals, the noise between the spectral harmonics causes a distinct roughness of the processed signals. Therefore, by cascading different combinations of prototype filter to produce more no of sub bands and amplitude shaper so that the speech signals can be improved to reduce the noise, where as the original first-stage filter-bank system can still be used for compression and amplification. We also consider the algorithmic delay added when implementing filter-bank. This first stage can be applied to lower frequency bands only, so that the harmonics of speech especially for low content can be resolved. This paper presents a systematic description of the cascade filter-bank stage, discusses its influence on the processed signals in detail and further presents the results which indicate the improved performance of signal to noise ratio, computational complexity compared to the original single-stage filter-bank system.