Source-Normalized LDA for Robust Speaker Recognition Using i-Vectors From Multiple Speech Sources.
Mitchell McLarenDavid A. van LeeuwenPublished in: IEEE Trans. Speech Audio Process. (2012)
Keyphrases
- multiple sources
- speech recognition
- audio visual
- linear discriminant analysis
- speaker identification
- speaker diarization
- discriminant analysis
- speaker recognition
- speaker verification
- face recognition
- information sources
- automatic speech recognition
- single source
- speaker dependent
- speech signal
- noisy environments
- broadcast news
- language model
- probabilistic model
- feature vectors
- feature extraction