The second multi-channel multi-party meeting transcription challenge (M2MeT) 2.0): A benchmark for speaker-attributed ASR.
Yuhao LiangMohan ShiFan YuYangze LiShiliang ZhangZhihao DuQian ChenLei XieYanmin QianJian WuZhuo ChenKong Aik LeeZhijie YanHui BuPublished in: CoRR (2023)
Keyphrases
- multi party
- multi channel
- automatic speech recognition
- speech recognition
- privacy preserving
- automatic transcription
- spontaneous speech
- anti aliasing
- single channel
- handwriting recognition
- speech signal
- channel assignment
- hidden markov models
- word error rate
- pattern recognition
- speaker diarization
- mental states
- speaker verification
- virtual humans
- language model
- data broadcast
- description language
- domain independent