Improving Robustness of Deep Neural Network Acoustic Models via Speech Separation and Joint Adaptive Training.
Arun NarayananDeLiang WangPublished in: IEEE ACM Trans. Audio Speech Lang. Process. (2015)
Keyphrases
- acoustic models
- speech recognition
- neural network
- hidden markov models
- training process
- broadcast news
- automatic speech recognition
- speech recognizer
- speaker independent
- discriminative training
- pattern recognition
- speech signal
- training data
- language model
- training examples
- noisy environments
- speaker identification
- generative model
- supervised learning
- feature vectors
- machine learning