Adapt to MMIF version spec-0.2.1

marcverhagen commented 3 years ago

Some things remain to be done:

Find the TextDocument created by Kaldi and use that instead of the transcript in the documents list.
Find the alignments using the new scheme with TimeFrame, Token and Alignment types (it now still uses the old alignments of type "vanilla-forced-alignment".

Harder is how we want to deal with the named entities from the set of TextDocument created by Tesseract and that the NER then ran over. My first hunch is to take all those documents and glue them together. I would like to change the module so it is a bit more view-centric.

keighrim commented 3 years ago

For the second point, a new method get_alignments in mmif-python 0.2.2 can be helpful.

keighrim commented 1 year ago

This must have been fixed via #9 and #10. Or with the new work started by Hayden, this issue might be no longer relevant. @marcverhagen can you confirm?

marcverhagen commented 1 year ago

Maybe, I did notice a comment saying that I did not use get_alignments because it caused errors. SO I need to revisit this.

marcverhagen commented 1 year ago

Yes, this can be closed. The some-things-remaining-to-be-done were done.

clamsproject / mmif-visualizer

Adapt to MMIF version spec-0.2.1 #3