Multi-Frame GMM-Based Block Quantisation for Distributed Speech Recognition Under Noisy Conditions.
Stephen SoKuldip K. PaliwalPublished in: ICASSP (1) (2006)
Keyphrases
- speech recognition
- multi frame
- speaker identification
- noisy environments
- voice activity detection
- speaker recognition
- motion estimation
- gaussian mixture model
- speech processing
- hidden markov models
- language model
- point correspondences
- pattern recognition
- super resolution
- speech signal
- image sequences
- automatic speech recognition
- speech synthesis
- speech recognition systems
- speaker dependent
- speech recognizer
- motion analysis
- optic flow
- computationally efficient
- optical flow
- speech recognition technology
- video sequences
- filtering algorithm
- speaker verification
- mixture model
- speaker independent
- computer vision
- signal to noise ratio
- expectation maximization
- mel frequency cepstral coefficients
- multiscale
- image processing