An investigation on DNN-derived bottleneck features for GMM-HMM based robust speech recognition.
Yongbin YouYanmin QianTianxing HeKai YuPublished in: ChinaSIP (2015)
Keyphrases
- speech recognition
- hidden markov models
- speaker identification
- speech recognition systems
- mel frequency cepstral coefficients
- noisy environments
- speech processing
- pattern recognition
- feature vectors
- automatic speech recognition
- speech signal
- feature extraction
- language model
- gaussian mixture model
- speaker recognition
- speech understanding
- cepstral coefficients
- speaker dependent
- voice activity detection
- speech recognizer
- speech synthesis
- probabilistic neural network
- feature set
- feature space
- speech recognition errors
- speech retrieval
- speech recognition technology
- speaker diarization
- speaker independent
- low level
- computer vision