Advancing the Dimensionality Reduction of Speaker Embeddings for Speaker Diarisation: Disentangling Noise and Informing Speech Activity.
You Jin KimHee-Soo HeoJee-Weon JungYoungki KwonBong-Jin LeeJoon Son ChungPublished in: ICASSP (2023)
Keyphrases
- dimensionality reduction
- speech recognition
- speaker verification
- speaker recognition
- audio visual
- automatic speech recognition
- noisy environments
- speaker identification
- speaker dependent
- prosodic features
- speaker diarization
- low dimensional
- speech signal
- high dimensional data
- speech synthesis
- vocal tract
- automatic speech recognition systems
- principal component analysis
- gaussian mixture model
- pattern recognition
- feature extraction
- synthesized speech
- speech enhancement
- automatic transcription
- noise reduction
- manifold learning
- noisy speech
- background noise
- speaker adaptation
- high dimensionality
- visual information
- broadcast news
- speech sounds
- acoustic features
- linear discriminant analysis
- speech recognizer
- hidden markov models
- high dimensional
- speaker independent
- data representation
- noisy data
- pattern recognition and machine learning
- audio stream
- computer vision
- acoustic models
- spontaneous speech
- multi modal
- vector space
- human activities
- noise level
- signal to noise ratio