SLOGD: Speaker Location Guided Deflation Approach to Speech Separation.
Sunit SivasankaranEmmanuel VincentDominique FohrPublished in: ICASSP (2020)
Keyphrases
- speech recognition
- speaker recognition
- audio visual
- automatic speech recognition
- speaker verification
- speaker identification
- prosodic features
- automatic speech recognition systems
- speaker dependent
- speech signal
- vocal tract
- speaker diarization
- hidden markov models
- multi modal
- automatic transcription
- speech sounds
- location information
- vector quantization
- text to speech
- speech recognizer
- audio features
- recognition engine
- synthesized speech
- noisy environments
- audio stream
- feature selection
- endpoint detection
- gaussian mixture model
- sound source
- dialogue system