Enhancing the complex-valued acoustic spectrograms in modulation domain for creating noise-robust features in speech recognition.
Hsin-Ju HsiehBerlin ChenJeih-Weih HungPublished in: APSIPA (2015)
Keyphrases
- speech recognition
- speech recognition systems
- noisy environments
- complex valued
- speech signal
- noisy speech
- mel frequency cepstral coefficients
- hidden markov models
- speech synthesis
- language model
- pattern recognition
- speaker independent
- automatic speech recognition
- real valued
- speaker identification
- speech recognizers
- singular spectrum analysis
- acoustic models
- cepstral coefficients
- speech recognizer
- background noise
- classification accuracy
- noise reduction
- feature set
- speech enhancement
- blind source separation
- computer vision
- maximum likelihood
- visual speech
- recurrent neural networks
- feature vectors
- feature extraction
- speech retrieval
- word recognition
- machine learning
- data mining