Improving wideband speech recognition using mixed-bandwidth training data in CD-DNN-HMM.
Jinyu LiDong YuJui-Ting HuangYifan GongPublished in: SLT (2012)
Keyphrases
- speech signal
- training data
- speech recognition
- hidden markov models
- training process
- automatic speech recognition
- automatic speech recognition systems
- acoustic models
- multi stream
- vocal tract
- speech quality
- non stationary
- speaker identification
- noisy environments
- learning algorithm
- classification accuracy
- supervised learning
- decision trees
- data sets
- speech synthesis
- prior knowledge
- speech processing
- recognition engine
- test data
- test set
- audio visual
- keyword spotting
- phoneme recognition
- classification models
- class labels
- training set
- speech recognizer
- video streaming
- handwritten word recognition
- speech enhancement
- signal to noise ratio
- support vector machine
- neural network