Directional ASR: A New Paradigm for E2E Multi-Speaker Speech Recognition with Source Localization.
Aswin Shanmugam SubramanianChao WengShinji WatanabeMeng YuYong XuShi-Xiong ZhangDong YuPublished in: ICASSP (2021)
Keyphrases
- speech recognition
- source localization
- automatic speech recognition
- speech signal
- sound source
- hidden markov models
- language model
- speaker identification
- wireless sensor networks
- speaker dependent
- speech synthesis
- noisy environments
- speech recognition technology
- speech processing
- speaker independent
- speech retrieval
- speaker adaptation
- speech recognizer
- pattern recognition
- word error rate
- speech recognizers
- speaker recognition
- broadcast news
- speech recognition systems
- word recognition
- vocal tract
- acoustic features
- acoustic models
- handwriting recognition
- probabilistic model
- neural network