Dimensionality reduction of large TDOA vectors for speaker diarization.
Deepu VijayasenanFabio ValentePublished in: SAPA@INTERSPEECH (2012)
Keyphrases
- speaker diarization
- dimensionality reduction
- principal components
- speech recognition
- high dimensional
- feature extraction
- principal component analysis
- low dimensional
- pattern recognition
- high dimensional data
- feature vectors
- broadcast news
- data points
- bayesian information criterion
- audio stream
- speaker verification
- linear discriminant analysis
- dimension reduction
- feature selection
- computer vision