ACL RD-TEC 1.0 Summarization of W97-0117
Paper Title:
A NATURAL LANGUAGE CORRECTION MODEL FOR CONTINUOUS SPEECH RECOGNITION
A NATURAL LANGUAGE CORRECTION MODEL FOR CONTINUOUS SPEECH RECOGNITION
Authors: Tomek Strzalkowski and Ronald Brandow
Primarily assigned technology terms:
- advanced speech recognition
- algorithm
- asr system
- asr transcription
- automated speech recognition
- c-box approach
- continuous speech recognition
- database
- error-driven learning
- hmms
- language correction
- language processing
- learning
- learning techniques
- machine translation
- medical dictation
- modeling
- modeling technique
- n-best alternative transcription
- n-gram weighting
- natural language processing
- optimization
- parallel training
- parser
- part-of-speech tagging
- post-processing
- postprocessing
- probabilistie modeling
- probabilistie modeling technique
- processing
- ranking
- re-ranking
- recognition
- recognition system
- recognition systems
- repair
- rule generalization
- rule interference
- rule validation
- speech production
- speech recognitio n
- speech recognition
- speech recognition system
- speech recognition systems
- speech system
- speech transcription
- speech understanding
- spelling
- spelling correction
- syntactic parser
- tagging
- terminology
- text alignment
- training process
- transcription
- validation
- weighting
- weighting method
Other assigned terms:
- american english
- approach
- asr output
- baseline performance
- broadcast news
- context features
- context-free rule
- continuous speech
- correlation
- dialog
- distribution
- error rate
- evaluations
- experimental results
- fact
- feature
- healthcare
- heuristics
- human intervention
- hypotheses
- language model
- language models
- learnability
- lexicon
- lexieal well-formedness
- likelihood
- linguistic
- linguistic information
- medical terminology
- method
- n-gram
- names
- natural language
- paragraphs
- parallel text
- part-of-speech
- part-of-speech information
- perplexity
- process
- recognition accuracy
- rule set
- sentence
- sentences
- sparse data
- stress
- style
- sublanguage
- technique
- technology
- text
- text corpus
- training
- training corpus
- training data
- training phase
- training samples
- training text
- transcriptions
- trigram
- umls
- understanding
- vocabulary
- voice input
- word
- word error rate
- word error rates
- word level
- words
