Multi-Task Transformer with Input Feature Reconstruction for Dysarthric Speech Recognition.
Chaoyue DingShiliang SunJing ZhaoPublished in: ICASSP (2021)
Keyphrases
- speech recognition
- multi task
- multi task learning
- hidden markov models
- learning tasks
- language model
- cepstral coefficients
- speech signal
- speech recognition systems
- feature selection
- speech recognizer
- speech synthesis
- automatic speech recognition
- gaussian processes
- multi class
- feature set
- pattern recognition
- speaker identification
- feature vectors
- speaker independent
- transfer learning
- feature subset
- learning problems
- image features
- feature extraction
- natural language processing
- information extraction
- learning process
- e learning