Cepstral smoothing of separated signals for underdetermined speech separation.
Yumi AnsaShoko ArakiShoji MakinoTomohiro NakataniTakeshi YamadaAtsushi NakamuraNobuhiko KitawakiPublished in: ISCAS (2010)
Keyphrases
- cepstral features
- source signals
- speech recognition
- blind source separation
- blind separation
- speech signal
- audio features
- independent component analysis
- source separation
- linear prediction
- mixing matrix
- sound source
- blind signal separation
- acoustic signals
- fundamental frequency
- automatic speech recognition
- audio signals
- speech synthesis
- noisy environments
- signal processing
- hidden markov models
- feature extraction
- language model
- gaussian white noise
- speaker identification
- spectral analysis
- audio visual
- speaker verification
- blind deconvolution
- smoothing methods
- recognition engine
- smoothing algorithm
- independent components
- nonlinear functions
- non stationary
- single channel
- text to speech
- spoken language
- human computer interaction
- acoustic signal
- visual features
- principal component analysis
- multiresolution
- multiscale