Esophageal Speech enhancement using modified voicing source

Abstract

This paper presented the modification to Esophageal Speech (ES) enhancement using Adaptive Gain Equalizer (AGE) for modifying the voicing source. However, the voicing source used previously with AGE, obtained using conventional Linear Predication (LP) vocal tract transfer function (AGE-LP), has produced low quality speech due to sensitivity to background noise. The better quality ES can be obtained by estimating voicing source through Iterative Adaptive Inverse Filtering utilized Weighted Linear Prediction (WLP) vocal tract transfer function (AGE-IAIF). The system performance evaluated through Harmonic to Noise Ratio (HNR), and system has shown 3 dB enhancement by AGE-IAIF over previously enhancement method AGE-LP.