SememeASR: Boosting Performance of End-to-End Speech Recognition against Domain and Long-Tailed Data Shift with Sememe Semantic Knowledge.
Jiaxu ZhuChanghe SongZhiyong WuHelen MengPublished in: INTERSPEECH (2023)
Keyphrases
- end to end
- speech recognition
- semantic knowledge
- training data
- pattern recognition
- domain specific
- domain experts
- speech signal
- prior knowledge
- knowledge discovery
- speech synthesis
- neural network
- domain ontology
- semantic information
- language model
- hidden markov models
- feature extraction
- artificial intelligence
- machine learning