Single channel speech music separation using nonnegative matrix factorization and spectral masks.
Emad M. GraisHakan ErdoganPublished in: DSP (2011)
Keyphrases
- nonnegative matrix factorization
- single channel
- sound source
- speech enhancement
- audio features
- speech signal
- multi channel
- linear prediction
- audio visual
- matrix factorization
- data representation
- negative matrix factorization
- spectral clustering
- least squares
- speech recognition
- independent component analysis
- principal component analysis
- document clustering
- frequency domain
- vocal tract
- original data
- edge detection
- wiener filter
- objective function
- wavelet decomposition
- machine learning
- denoising
- semi supervised
- collaborative filtering
- brain activity
- image data