Open joshkyh opened 3 years ago
Hi Joshua! I normalized the volumes using pydub. The result looks reasonable to me.
I have already uploaded the narrative_fine results of BMI and GS in the database.
Run vp on entire earnings call set. #9 Also run v on entires earnings call set.
(1) Please check whether the 32000 harded variable changes could improve the quality of the tone annotations. (2) If there are too many angers, it is likely that the volume is too loud for the model, hence we might require normalizing volumes with ffmpeg.
If the above is satisfactory, build a new py file to upload the stage3/narrative_fine.csv