Combining Conformer and Dual-Path-Transformer Networks for Single Channel Noisy Reverberant Speech Separation.
William RavenscroftStefan GoetzeThomas HainPublished in: ICASSP (2024)
Keyphrases
- single channel
- speech signal
- speech enhancement
- sound source
- noisy environments
- multi channel
- speech recognition
- automatic speech recognition
- vocal tract
- noise reduction
- additive noise
- frequency domain
- wiener filter
- linear prediction
- synthetic and real images
- pattern recognition
- audio visual
- independent component analysis
- non stationary