MFCCA: Multi-Frame Cross-Channel attention for multi-speaker ASR in Multi-party meeting scenario.
Fan YuShiliang ZhangPengcheng GuoYuhao LiangZhihao DuYuxiao LinLei XiePublished in: CoRR (2022)
Keyphrases
- multi party
- multi frame
- cross channel
- automatic speech recognition
- speech recognition
- privacy preserving
- point correspondences
- motion estimation
- image sequences
- super resolution
- computationally efficient
- mental states
- free riding
- motion analysis
- description language
- video frames
- optic flow
- optical flow
- filtering algorithm
- distributed systems
- hidden markov models
- single channel
- flow field
- service delivery
- image processing