Télécom Paris

Eigendomain-based noise estimation with the minimum statistics approach
  • Vinesh Bhunjun, (Dept of Elec Eng, Imperial College London)
  • Mike Brookes, (Dept of Elec Eng, Imperial College London)
  • Jimi Y. C. Wen, (Dept of Elec Eng, Imperial College London)
  • Noise and acoustic environments and characteristics
  • Sound enhancement and sound separation
  • Noise reduction techniques
Get the paper in PDF format
Acrobat Reader (version 5 minimum) is necessary to read this document.

One of the challenges for single-channel speech enhancement is to estimate the noise statistics from a signal containing both speech and noise. In this paper, we present a technique for eigendomain-based noise estimation that uses minimum statistics to control the adaptation rate along each eigenvector. We demonstrate that this technique gives robust noise tracking for non-stationary noise.

©2006 Télécom Paris/TSI
Edition : Télécom Paris -- 2006