Simplifying long short-term memory acoustic models for fast training and decoding.
Yajie MiaoJinyu LiYongqiang WangShi-Xiong ZhangYifan GongPublished in: ICASSP (2016)
Keyphrases
- acoustic models
- speech recognition
- hidden markov models
- automatic speech recognition
- discriminative training
- training process
- recurrent neural networks
- speech recognizer
- training data
- supervised learning
- maximum likelihood
- speaker independent
- data sets
- fuzzy logic
- training set
- pattern recognition
- learning algorithm
- genetic algorithm