Efficient use of DNN bottleneck features in generalized variable parameter HMMs for noise robust speech recognition.
Rongfeng SuXurong XieXunying LiuLan WangPublished in: INTERSPEECH (2015)
Keyphrases
- speech recognition
- noisy environments
- hidden markov models
- speech recognition systems
- speech signal
- cepstral coefficients
- pattern recognition
- language model
- automatic speech recognition
- speech processing
- speech synthesis
- speech recognizer
- speaker independent
- feature set
- digit recognition
- speaker identification
- noisy speech
- speech enhancement
- speech retrieval
- feature extraction
- feature vectors
- background noise
- handwriting recognition
- noise reduction
- classification accuracy
- neural network
- mel frequency cepstral coefficients
- extracted features
- signal to noise ratio
- speaker adaptation
- speech recognition errors