Multi-task Learning with Augmentation Strategy for Acoustic-to-word Attention-based Encoder-decoder Speech Recognition.
Takafumi MoriyaSei UenoYusuke ShinoharaMarc DelcroixYoshikazu YamaguchiYushi AonoPublished in: INTERSPEECH (2018)
Keyphrases
- speech recognition
- speech recognition systems
- multi task learning
- speech recognizers
- speech recognizer
- multi task
- learning tasks
- hidden markov models
- speaker independent
- language model
- gaussian processes
- learning problems
- high order
- automatic speech recognition
- theoretical analysis
- speech signal
- speech synthesis
- transfer learning
- noisy environments
- pattern recognition
- acoustic models
- speaker identification
- learning models
- bit rate
- acoustic features
- motion estimation
- feature selection