Lightweight Feature Encoder for Wake-Up Word Detection Based on Self-Supervised Speech Representation.
Hyungjun LimYounggwan KimKiho YeomEunjoo SeoHoodong LeeStanley Jungkyu ChoiHonglak LeePublished in: ICASSP (2023)
Keyphrases
- lightweight
- word counts
- feature representation
- development environments
- object detection
- co occurrence
- dos attacks
- speech signal
- communication infrastructure
- spoken document retrieval
- english text
- detection algorithm
- wireless sensor networks
- computational complexity
- power consumption
- rfid tags
- rate distortion
- speech recognition
- image representation
- lexical features
- speech recognition systems
- text input
- feature vectors