Unsupervised learning of time-frequency patches as a noise-robust representation of speech.
Maarten Van SegbroeckHugo Van hammePublished in: Speech Commun. (2009)
Keyphrases
- unsupervised learning
- noisy environments
- image noise
- speech recognition
- speech enhancement
- supervised learning
- natural images
- signal to noise ratio
- noise reduction
- salt pepper
- geometric distortions
- automatic speech recognition
- image classification
- object recognition
- image patches
- imaging artifacts
- greater robustness
- knn
- visual properties
- background noise
- semi supervised
- registration errors
- text classification
- expectation maximization
- audio visual
- estimation error
- feature descriptors
- reinforcement learning
- missing data
- noisy data