Discriminatively trained features using fMPE for multi-stream audio-visual speech recognition.
Jing HuangDaniel PoveyPublished in: INTERSPEECH (2005)
Keyphrases
- audio visual speech recognition
- multi stream
- discriminatively trained
- audio visual
- object detection
- hidden markov models
- low level
- feature set
- audio signal
- image features
- feature vectors
- high dimensional
- generative model
- extracted features
- discriminative learning
- multiscale
- co occurrence
- multimedia
- multi modal
- noisy environments
- feature space
- pattern recognition