Speech dereverberation by data-dependent beamforming with signal pre-whitening.
Thomas DietzenN. HuleihelA. SpriefWouter TirrySimon DocloMarc MoonenToon van WaterschootPublished in: EUSIPCO (2015)
Keyphrases
- data dependent
- speech enhancement
- linear prediction
- single channel
- frequency domain
- noisy environments
- speech signal
- speech recognition
- sound source
- direction of arrival
- rademacher complexity
- risk bounds
- generalization bounds
- audio visual
- lossless compression
- automatic speech recognition
- signal to noise ratio
- data sets
- energy functional
- hash functions
- prediction error
- non stationary
- reproducing kernel hilbert space
- noise reduction
- deformable models
- subband
- higher order
- denoising
- pairwise