Adapting Speech Separation to Real-World Meetings using Mixture Invariant Training.
Aswin SivaramanScott WisdomHakan ErdoganJohn R. HersheyPublished in: ICASSP (2022)
Keyphrases
- real world
- hearing impaired
- audio visual
- wide range
- speaker diarization
- case study
- synthetic data
- test set
- speech recognition
- training set
- training phase
- training samples
- data sets
- data mining
- blind separation
- gaussian distribution
- affine transformation
- training process
- automatic transcription
- affine invariant
- online learning
- speech signal
- supervised learning
- automatic speech recognition
- broadcast news
- text to speech
- probabilistic model
- hidden markov models
- bayesian networks