ACL RD-TEC 1.0 Summarization of P06-1140
Paper Title:
LEARNING TO SAY IT WELL: RERANKING REALIZATIONS BY PREDICTED SYNTHESIS QUALITY
LEARNING TO SAY IT WELL: RERANKING REALIZATIONS BY PREDICTED SYNTHESIS QUALITY
Authors: Crystal Nakatsu and Michael White
Primarily assigned technology terms:
- algorithm
- classification
- computational linguistics
- cross-validation
- database
- dialogue system
- discriminative reranking
- discriminative training
- festival speech synthesis
- finite-state transducers
- instance-based surface realization
- language dialogue system
- language generation
- language generation and synthesis
- language generator
- language modeling
- learning
- markup language
- modeling
- multimodal dialogue
- openccg realizer
- optimization
- parsing
- phrasing
- planner
- prosody prediction
- rankboost
- ranking
- rating
- reading
- reranking
- scoring
- search
- selection algorithm
- selection search
- sentence planner
- sentence planning
- speech synthesis
- speech synthesis system
- speech synthesizer
- spoken language dialogue
- support vector machine
- surface realization
- svm ranking
- synthesis
- synthesis system
- synthesiser
- synthesizer
- text analysis
- textto-speech
- transducers
- unit selection
- unit selection algorithm
- unit selection search
- weighted finite-state transducers
Other assigned terms:
- 10-fold cross-validation
- accent
- approach
- association for computational linguistics
- backoff
- bigram
- case
- coherence
- correlation
- dialogue context
- dialogue history
- dialogues
- diphone coverage
- evaluations
- fact
- feature
- feature set
- feature sets
- feature vector
- feature vectors
- generation
- gesture
- grammar
- handwriting
- hypothesis
- information structure
- intelligibility
- intonation
- intonational phrase
- knowledge
- labeling
- language model
- lattice
- lattices
- linguistics
- local coherence
- logical forms
- markup
- meaning
- method
- n-gram
- n-gram model
- n-gram models
- names
- optimization problem
- paraphrase
- paraphrases
- pauses
- phrase
- pitch
- pitch accent
- priori
- probabilities
- procedure
- proper names
- prosody
- segments
- semantic
- semantic classes
- sentence
- sentence structure
- sentences
- speech corpus
- speech database
- spoken language
- standard deviation
- style
- support vector
- syntactic structure
- synthesis quality
- synthesized speech
- synthetic speech
- test data
- testing set
- text
- text quality
- training
- training data
- trigram
- trigram model
- triphone
- unigram
- user
- user model
- utterance
- word
- word lattice
- word lattices
- word sequences
- words
