Combination of MVDR beamforming and single-channel spectral processing for enhancing noisy and reverberant speech.
Benjamin CauchiIna KodrasiRobert RehrStephan GerlachAnte JukicTimo GerkmannSimon DocloStefan GoetzePublished in: EURASIP J. Adv. Signal Process. (2015)
Keyphrases
- speech enhancement
- speech signal
- single channel
- noisy environments
- linear prediction
- multi channel
- speech recognition
- sound source
- minimum variance
- vocal tract
- automatic speech recognition
- noise reduction
- hidden markov models
- signal to noise ratio
- non stationary
- additive noise
- wiener filter
- independent component analysis
- blind source separation
- frequency domain
- brain activity
- video sequences