Given a youtube video id and language, aligns audio to transcript using the Montreal Forced Aligner. Before using, must have MFA downloaded and installed. https://montreal-forced-aligner.readthedocs.io/en/latest/introduction.html
For the language you'd like to align to, place the language and g2p models (.zip files) in the working directory
https://montreal-forced-aligner.readthedocs.io/en/latest/pretrained_models.html
and change the G2P_MODEL
and LANG_MODEL
paramters to their locations. Additionally, change the MFA_ALIGNER
parameter to the directory where you downloaded the Montreal Forced Aligner.
Example: To align the video Lga-oupKJx4 (a spanish math video), place a trained spanish model, spanish g2p model, and the MFA in the same directory as align.py. Then run
python align.py Lga-oupKJx4 es