Word level confidence measures using n-best sub-hypotheses likelihood ratio.
Beng Tiong TanYong GuTrevor ThomasPublished in: INTERSPEECH (2001)
Keyphrases
- likelihood ratio
- word level
- confidence measures
- hypothesis testing
- ground truth
- stereo vision
- confidence measure
- language independent
- document images
- hypothesis test
- document analysis
- optical flow
- machine translation
- n gram
- character recognition
- word segmentation
- sentence level
- viterbi algorithm
- image analysis
- semantic roles
- high quality
- word recognition
- statistical tests
- stereo matching
- vision system
- language model
- training data