Surround noise cancellation and speech enhancement using Sub Band filtering and Spectral Subtraction

Abstract

Background/Objectives: To propose an algorithm based on Sub Band filtering, Spectral Subtraction, and Voice Activity Detector (VAD) for surround noise cancellation and speech enhancement. Methods/Statistical Analysis: Simulation study has been carried out to test the performance of the algorithm is tested for two real time speech signal (both male and female), considering three different surround noise sources such as fan, AC duct and Computer at various SNR dB levels. Findings: Simulation results show that the algorithm works efficiently in reducing the surround noise, which when mixed in a clean speech signal. For male speech signal corrupted by noise from fan, has -55 dB reduction of noise level, whereas for the female speech signal corrupted by noise from AC duct, the noise level is reduced to -65 dB and for Male speech signal corrupted by noise from computer, the noise level is reduce to -75 dB. Conclusions/Improvements: Presently the algorithm is proposed for stationary sound produced in a room, mainly for teleconferencing application, but the algorithm can also be tested for musical noise. Algorithm can be extended for sudden non stationary noises.