Transforming features to compensate speech recogniser models for noise.
Rogier C. van DalenFederico FlegoMark J. F. GalesPublished in: INTERSPEECH (2009)
Keyphrases
- noisy environments
- predictive power
- image features
- prior knowledge
- feature vectors
- feature extraction
- speech recognition
- image statistics
- learned models
- extracted features
- noisy data
- co occurrence
- machine learning algorithms
- neural network
- statistical models
- generative model
- learning models
- machine learning approaches
- facial expressions
- classification accuracy
- probabilistic model
- abstraction levels
- bayesian networks
- text to speech