Investigation of End-to-End Speaker-Attributed ASR for Continuous Multi-Talker Recordings.
Naoyuki KandaXuankai ChangYashesh GaurXiaofei WangZhong MengZhuo ChenTakuya YoshiokaPublished in: SLT (2021)
Keyphrases
- end to end
- automatic speech recognition
- speech recognition
- spontaneous speech
- acoustic features
- audio visual
- wireless ad hoc networks
- admission control
- congestion control
- multipath
- speech signal
- speaker verification
- high bandwidth
- ad hoc networks
- noisy environments
- rate allocation
- language model
- content delivery
- broadcast news
- sound source
- hidden markov models
- internet protocol
- motion estimation