DOA estimation of speech signals using semi-blind source separation techniques.
Ilyas PotamitisPanagiotis ZervasNikos FakotakisPublished in: INTERSPEECH (2004)
Keyphrases
- blind source separation
- speech signal
- doa estimation
- sound source
- speech recognition
- blind separation
- non stationary
- canonical correlation analysis
- independent component analysis
- hidden markov models
- eigenvalue decomposition
- convolutive mixtures
- signal to noise ratio
- single channel
- source signals
- pattern recognition
- neural network
- visual features
- language model
- audio features
- least squares
- machine learning