Open keighrim opened 1 month ago
Since the text segmentation (lines) used in the sync annotation data doesn't exactly match the speaker turns, we need some additional steps to decide the turn boundaries within the text lines where two speakers' utterances are mixed/overlapping. A few ideas;
[ A;[you jim] B:[I am good] ]
<< mark the whole as B
(B's token is more than A's token) For 50-50 situation? Could do some arbitrary assigned, like total random assignment, always the first, etc.
New Feature Summary
It seems that we could relatively easily create some gold evaluation data for SD problem by combining the time-sync annotation and the speaker turn markers in our "gold" transcript files.
Related
There's the "cleaner" code that removes the speaker markers (https://github.com/clamsproject/clams-utils/issues/2), and we should be able to "reverse" the functionality to obtain the speaker markers, to associate with the time frames for series of their utterances.
Alternatives
No response
Additional context
No response