Unsupervised Learning of Overlapped Speech Model Parameters For Multichannel Speech Activity Detection in Meetings.
Kornel LaskowskiTanja SchultzPublished in: ICASSP (1) (2006)
Keyphrases
- speaker diarization
- unsupervised learning
- speech activity detection
- speech recognition
- broadcast news
- supervised learning
- bayesian information criterion
- speaker identification
- semi supervised
- dimensionality reduction
- mixture model
- expectation maximization
- text classification
- machine learning
- model selection
- object recognition
- gaussian mixture model
- automatic speech recognition
- speaker verification
- feature selection
- audio visual
- emotion recognition
- hidden markov models
- feature space