Improving speech recognition in reverberation using a room-aware deep neural network and multi-task learning.
Ritwik GiriMichael L. SeltzerJasha DroppoDong YuPublished in: ICASSP (2015)
Keyphrases
- speech recognition
- multi task learning
- neural network
- multi task
- pattern recognition
- multiple tasks
- learning tasks
- hidden markov models
- gaussian processes
- language model
- speech signal
- automatic speech recognition
- learning problems
- high order
- speech recognizer
- theoretical analysis
- speech synthesis
- learning algorithm
- transfer learning
- sound source
- knn
- learning experience
- image classification
- higher order
- semi supervised
- speaker identification