ASR Posterior-Based Loss for Multi-Task End-to-End Speech Translation.
Yuka KoKatsuhito SudohSakriani SaktiSatoshi NakamuraPublished in: Interspeech (2021)
Keyphrases
- end to end
- multi task
- automatic speech recognition
- speech recognition
- speech signal
- multi task learning
- learning tasks
- multi class
- gaussian processes
- hidden markov models
- multiple tasks
- feature selection
- sparse learning
- transfer learning
- learning problems
- congestion control
- probabilistic model
- real world
- text localization and recognition
- error propagation
- gaussian process
- information gain
- bayesian framework
- language model
- training data