Speech Separation Methodology for Hearing Aid

In the design of hearing aids (HA), real-time speech-enhancement is done. The digital should provide high signal-to-noise ratio, gain improvement and eliminate feedback. generic performance towards different frequencies varies non uniform. Existing noise cancellation speech separation methods drops voice magnitude under environment. HA for frequency response suppression reduce required signal strength also. So, uniform sub band analysis poor when aid concern. this paper, a method using Non-negative Matrix Factorization (NMF) algorithm proposed wavelet decomposition. Proposed non-uniform filter-bank was validated by parameters like power, Signal-to-noise ratio (SNR), Mean Square Error (MSE), Signal to Noise Distortion Ratio (SINAD), Spurious-free dynamic range (SFDR), error time. recordings before after evaluated quality objective measures International Telecommunication Union -Telecommunication standard ITU-T P.862.