Leveraging Pretrained ASR Encoders for Effective and Efficient End-to-End Speech Intent Classification and Slot Filling.
He HuangJagadeesh BalamBoris GinsburgPublished in: CoRR (2023)
Keyphrases
- end to end
- speech recognition
- automatic speech recognition
- text localization and recognition
- multipath
- admission control
- ad hoc networks
- feature vectors
- high bandwidth
- wireless ad hoc networks
- content delivery
- pattern recognition
- congestion control
- speech signal
- spontaneous speech
- transport layer
- real time
- real world
- rate allocation
- computational complexity
- scalable video
- image quality