LAAC-LSCP / ChildProject

Python package for the management of day-long recordings of children.
https://childproject.readthedocs.io
MIT License
13 stars 5 forks source link

Adding VTC annotations to eaf-builder #333

Closed marianne-m closed 2 years ago

marianne-m commented 2 years ago

Is your feature request related to a problem? Please describe. The eaf-builder doesn't have an option to import the VTC annotations. This option would simplify and speed up the annotation work for more complex tasks like VCM or MSC. Instead of annotating the segment AND do the second task, the annotator would only have to correct the VTC annotation (if necessary) and could focus more on the second task.

Describe the solution you'd like Adding a --import-vtc-annotations option to eaf-builder

lucasgautheron commented 2 years ago

Just thinking about this more generally, maybe we should have an option to pre-fill the eaf from any kind of annotation ?

alecristia commented 2 years ago

tagging this to talk it over tomorrow!

lucasgautheron commented 2 years ago

Alright! Just did some un-tested, very preliminary work here: #334

lucasgautheron commented 2 years ago

Ok, it seems to be working (kind of), maybe we want to change how speaker_id is assigned based on the VTC's speaker_type.

But this implementation should work for VTC annotations, and other eaf annotations. Not for the LENA but it should be easy to make it work for it too (just need to infer the tier based on speaker_type rather than speaker_id).

lucasgautheron commented 2 years ago

Should now be working with LENA annotations too!