MIMO Self-Attentive RNN Beamformer for Multi-Speaker Speech Separation.
Xiyun LiYong XuMeng YuShi-Xiong ZhangJiaming XuBo XuDong YuPublished in: Interspeech (2021)
Keyphrases
- speech recognition
- audio visual
- speaker recognition
- automatic speech recognition
- speaker verification
- speaker identification
- recurrent neural networks
- nearest neighbor
- prosodic features
- speaker diarization
- speech signal
- speaker dependent
- vocal tract
- automatic speech recognition systems
- multipath
- minimum variance
- signal to noise ratio
- automatic transcription
- speech recognizer
- mimo systems
- spontaneous speech
- non stationary
- synthesized speech
- speech sounds
- gaussian mixture model
- visual attention
- pre attentive
- broadcast news
- acoustic features
- linear prediction
- fading channels