A perceptual signal subspace approach for speech enhancement in colored noise.
Firas JablounBenoît ChampagnePublished in: ICASSP (2002)
Keyphrases
- speech enhancement
- signal subspace
- noisy environments
- noise reduction
- signal to noise ratio
- single channel
- correlation matrix
- linear array
- speech signal
- sensor array
- background noise
- hyperspectral
- linear prediction
- additive noise
- estimation error
- wiener filter
- multi channel
- speech recognition
- synthetic aperture radar
- vocal tract
- human visual system
- multiscale
- edge preserving
- eigenvalue decomposition
- pattern recognition