Denoising Esophageal Speech using Combination of Complex and Discrete Wavelet Transform with Wiener filter and Time Dilated Fourier Cepstra

Abstract

Esophageal speech is one of the pathological voices, which is known to be weak in intelligibility and hard to understand. Our approach's main idea is to reduce the esophageal speech noises using two-hybrid methods. This paper aims to merge the advantages of wavelet-based methods such as DWT and DTCWT, along with the standard methods such as the Wiener filter and the time dilated Fourier. The first hybrid method applies the filters on the vocal tract cepstrum, while the second one applies them at the synthesis stage. Two experiments were conducted as well to evaluate the results by objective analysis. The results obtained by the proposed hybrid methods gave good performances.