Switching Variational Auto-Encoders for Noise-Agnostic Audio-visual Speech Enhancement.
Mostafa SadeghiXavier Alameda-PinedaPublished in: CoRR (2021)
Keyphrases
- audio visual
- speech enhancement
- noisy environments
- noise reduction
- signal to noise ratio
- multi modal
- single channel
- speech signal
- background noise
- visual data
- sound source
- visual information
- wiener filter
- additive noise
- linear prediction
- multi stream
- smoothing algorithm
- speech recognition
- vocal tract
- multimedia
- audio features
- image segmentation
- space time
- independent component analysis
- multiscale