|
I'm working with some literature scholars, and after seeing me do parsing and NER on their texts, they've been asking about prosody and rhyme. I'm completely ignorant of this area, other than knowing it's possible for spoken text. Google Scholaring aside, where/who in the NLP literature should I start reading for a crash course on the subject? Are there off-the-shelf algorithms for labeling it in free text? |
|
If I understand correctly, you're looking for techniques to process text predicting the prosodic qualities that it would have if it were read aloud. This is typically an operation of a speech synthesis front end. Festival includes a prosody model that you should be able to use in a more-or-less off the shelf way. You'll probably have to hack into it to isolate the prosody model, but it shouldn't be too terrible. Julia Hirschberg did work on this while at AT&T. Mari Ostendorf and her students have looked at this as well. |
