Improving speech recognition of two simultaneous speech signals by integrating ICA BSS and automatic missing feature mask generation.
Ryu TakedaShun'ichi YamamotoKazunori KomataniTetsuya OgataHiroshi G. OkunoPublished in: INTERSPEECH (2006)
Keyphrases
- speech signal
- speech recognition
- blind source separation
- independent component analysis
- cepstral coefficients
- automatic speech recognition
- speaker identification
- noisy environments
- blind separation
- hidden markov models
- linear prediction
- background noise
- language model
- vocal tract
- speech recognizer
- spectral analysis
- speech synthesis
- pattern recognition
- spectral subtraction
- source signals
- non stationary
- speaker recognition
- multi modal
- acoustic features
- principal component analysis
- speech recognition systems
- neural network