Analysis of system identification and modified application to two - microphone speech enhancement

This paper provides analysis of identification of acoustic transfer functions between two microphones and investigation of modified application from two-microphone adaptive noise cancelling and beamforming methods. Based on this, we will perform real-time performance comparisons to obtain the best solution to speech enhancement and noise cancellation. Experiments are processed by software implementation using LabVIEW in a real environment, which is typical indoor office with moderate reverberation condition. The speech performances are analyzed in time and frequency domains using both stationary and nonstationary noises. The analysis on the three type of microphones configuration and computational complexity on NLMS algorithm and TDOA function have also been investigated, which could give rise to fundamental basis for further real-time applications using two-microphone as well as hardware prototype implementation of digital adaptive hearing aids. Keywords— ANC, beamforming, NLMS, system identification, TDOA, VAD