ubclaunchpad / minutes

:telescope: Speaker diarization via transfer learning

27 stars 5 forks source link

Closed chadlagore closed 7 years ago

chadlagore commented 7 years ago

TODO

Make a little CLI to automate collection.

Flow:

Prompt for video ID
Download audio and display some XML.
Prompt user for some delimiter information and interior (see extract_labels.py).
Cache this info in <YouTube_ID>/<YouTube_ID>.json
Run pipeline.py <YouTube_ID> to generate the labeled data.