ACL RD-TEC 1.0 Summarization of A00-2029
Paper Title:
PREDICTING AUTOMATIC SPEECH RECOGNITION PERFORMANCE USING PROSODIC CUES
PREDICTING AUTOMATIC SPEECH RECOGNITION PERFORMANCE USING PROSODIC CUES
Authors: Diane J. Litman and Julia B. Hirschberg and Marc Swerts
Primarily assigned technology terms:
- 25-fold cross-validation
- algorithm
- asr system
- automatic speech recognition
- classification
- classifier
- corpus analysis
- corpus collection
- cross-validation
- dialogue manager
- dialogue systems
- error handling
- finite-state dialogue
- greedy search
- hidden markov
- hidden markov model
- identification
- labeler
- language engineering
- learning
- learning algorithm
- learning program
- machine learning
- markov model
- one-to-one mapping
- post-processing
- predictor
- ranking
- recognition
- recognition technology
- recognizer
- search
- speech recognition
- speech recognition technology
- speech recognizer
- spoken dialogue
- spoken dialogue systems
- text-to-speech
- transcription
Other assigned terms:
- american english
- asr output
- classification model
- concept
- confidence score
- confidence scores
- dialogue state
- dialogue strategies
- dialogues
- duration
- error rate
- fact
- feature
- feature set
- feature sets
- grammar
- grammars
- hyperarticulation
- hypotheses
- hypothesis
- information gain
- interpretation
- lexicon
- mapping
- measure
- mechanisms
- message
- names
- pause
- pauses
- pitch
- predictive power
- process
- prosodic feature
- prosodic features
- prosody
- recognition accuracy
- silence
- speaking rate
- speaking style
- speech recognition performance
- style
- syllables
- technology
- terms
- training
- training data
- training example
- training material
- transcriptions
- user
- user behavior
- utterance
- word
- word error rate
- words
