A Closer Look at Wav2vec2 Embeddings for On-Device Single-Channel Speech Enhancement.
Ravi ShankarKe TanBuye XuAnurag KumarPublished in: ICASSP (2024)
Keyphrases
- speech enhancement
- single channel
- multi channel
- independent component analysis
- frequency domain
- wiener filter
- prior information
- signal to noise ratio
- noise reduction
- wavelet decomposition
- noisy environments
- linear prediction
- brain activity
- signal processing
- multiscale
- spatial domain
- sound source
- noisy images
- smoothing algorithm
- feature space