Academic Journals Database
Disseminating quality controlled scientific knowledge

A Family of Spectral Subtraction Algorithms for Tamil Speech Enhancement

Author(s): Vimala.C | V.Radha

Journal: International Journal of Soft Computing & Engineering
ISSN 2231-2307

Volume: 2;
Issue: 1;
Start page: 397;
Date: 2012;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Speech enhancement | Tamil Speech | Spectral Subtraction | Non linear Spectral Subtraction | MMSE | Log Spectral MMSE | SNR and MSE.

Speech enhancement aims to improve thespeech quality by using various techniques andalgorithms. Over the past several years there has beenconsiderable attention focused on the enhancement ofspeech degraded by several types of noise. Thedegradation of speech due to the presence of noise causessevere difficulties in various communicationenvironments. Noise suppression has numerousapplications like Human Computer Interaction,hands-free communications, Voice over IP (VoIP),hearing aids, teleconferencing system etc. For this issuethere is always a unique need for the technique whichoffers expected outcome with limited complexity inimplementation. Hence, in this paper a family of spectralsubtraction techniques is employed for Tamil speechnoise cancellation due to its simplicity. The algorithmsadopted for this research work are namely basic spectralsubtraction, Non linear Spectral Subtraction, MultiBandSpectral Subtraction, Minimum Mean Square Error(MMSE), and Log Spectral MMSE. All these algorithmsare analyzed and implemented for two types of noisesnamely white and babble noise. The performances ofthese algorithms are estimated based on SNR and MSEmeasures. Based on the experimental results, the Nonlinear spectral subtraction algorithm provides betterresults than any other adopted algorithms.
Why do you need a reservation system?      Affiliate Program