With over a year of ASMR archival data, I can generate some pretty interesting stats about YouTube content.
As of Feb 7, with 323 channels watched and 35,330 videos archived, a whopping 2,750 videos (7.8%) have been removed from YouTube.
A few more stats below
@sundaefloatz I haven't done much with transcribing, but I've looked into using techniques like Contrastive Language-Audio Pretraining (CLAP) for segment labeling, but I wasn't able to make it work well enough out of the box. At some point I'll try to fine-tune a model for ASMR specifically.
The channel I worked on for 6 years got deleted after 1 Strike. At the moment I don't know what to do. @TeamYouTube@YouTubeCreators I asked for your help, and I wake up to my life destroyed.
Finally got around to setting up my infrastructure to archive from rplay.
I was never a fan of that site because of how they intermingle SFW and NSFW videos. I mostly solved this via some pretty crazy regex to just save SFW ASMR :)