Non-linear maximum likelihood feature transformation for speech recognition.
Mohamed Kamal OmarMark Hasegawa-JohnsonPublished in: INTERSPEECH (2003)
Keyphrases
- speech recognition
- feature transformation
- speaker identification
- maximum likelihood
- speaker adaptation
- noisy environments
- gaussian mixture model
- speech signal
- hidden markov models
- language model
- speech synthesis
- automatic speech recognition
- em algorithm
- speaker independent
- speech recognizer
- pattern recognition
- linear classifiers
- feature extraction
- scale invariant
- expectation maximization
- speech recognition systems
- distance metric
- hyperparameters
- maximum a posteriori
- mixture model
- data mining
- feature space
- neural network
- training data
- face recognition
- image processing
- natural images
- natural language processing
- image analysis