Introducing the Turbo-Twin-HMM for Audio-Visual Speech Enhancement.
Steffen ZeilerHendrik MeutznerAhmed Hussen AbdelazizDorothea KolossaPublished in: INTERSPEECH (2016)
Keyphrases
- audio visual
- speech enhancement
- multi stream
- speech signal
- hidden markov models
- noisy environments
- noise reduction
- multi modal
- speech recognition
- single channel
- signal to noise ratio
- visual information
- vocal tract
- automatic speech recognition
- linear prediction
- multimedia
- visual data
- background noise
- smoothing algorithm
- sound source
- multi channel
- speaker identification
- wiener filter
- information retrieval
- multiscale
- audio features
- image processing
- edge detection
- pattern recognition
- three dimensional