On the robustness of linear discriminant analysis as a preprocessing step for noisy speech recognition.
Olivier SiohanPublished in: ICASSP (1995)
Keyphrases
- preprocessing step
- speech recognition
- linear discriminant analysis
- dimensionality reduction
- noisy environments
- principal component analysis
- feature extraction
- pattern recognition
- discriminant analysis
- preprocessing
- dimension reduction
- speech signal
- high dimensional data
- hidden markov models
- face recognition
- small sample size
- feature selection
- high dimensional
- automatic speech recognition
- speech synthesis
- feature space
- language model
- null space
- speech recognizer
- principal components analysis
- low dimensional
- support vector machine svm
- principal components
- locality preserving projections
- dimensionality reduction methods
- speech recognition systems
- high dimensionality
- unsupervised learning
- data points
- machine learning
- random projections
- neural network
- speaker identification
- object detection
- image processing
- supervised dimensionality reduction
- lower dimensional
- image classification
- computer vision
- scatter matrices