Improving deep neural networks based multi-accent Mandarin speech recognition using i-vectors and accent-specific top layer.
Mingming ChenZhanlei YangJizhong LiangYanpeng LiWenju LiuPublished in: INTERSPEECH (2015)
Keyphrases
- speech recognition
- spoken language
- neural network
- pattern recognition
- automatic speech recognition
- speech synthesis
- speech signal
- multi layer
- hidden markov models
- speech recognizer
- language model
- speaker independent
- genetic algorithm
- multiple layers
- broadcast news
- speech recognition systems
- language processing
- dialogue system
- domain specific
- recognition engine
- emotion recognition
- deep learning
- machine learning
- training process
- back propagation
- sufficient conditions
- video sequences
- multilayer perceptron
- neural network model
- application layer
- vector space
- speaker identification
- text to speech
- higher level
- multi modal
- fuzzy logic
- artificial neural networks
- prosodic features