Audiobook Dialogues as Training Data for Conversational Style Synthetic Voices.
Liisi PiitsHille PajupuuHeete SahkaiRene AltrovLiis ErmusKairi TamuriIndrek HeinMeelis MihklaIndrek KiisselEgert MännisaluKristjan SulusteJaan PajupuuPublished in: LREC (2022)
Keyphrases
- training data
- conversational agent
- multi party
- test data
- learning algorithm
- data sets
- training set
- training samples
- supervised learning
- classification accuracy
- real world
- multi modal
- labeled data
- domain knowledge
- natural language
- training dataset
- prior knowledge
- training examples
- noisy data
- learned from training data
- real images are presented
- speech acts
- dialogue system
- machine learning
- positive examples
- classification models
- generalization error
- speech recognition
- authorship attribution
- database
- unlabeled data