Is there a lightweight tool, with source code available, for generating subtitles from videos in Finnish/Swedish/Norwegian? (lemmy.dbzer0.com)

submitted 4 hours ago by flake@lemmy.dbzer0.com to c/opensource@lemmy.ml

5 comments fedilink hide all child comments

top 5 comments

sorted by: hot top controversial new old

[-] konki@lemmy.one 3 points 2 hours ago

https://ai.nb.no/models/

The National Library of Norway has an AI-lab that finetune some models; amomg them Whisper.

[-] SGforce@lemmy.ca 4 points 4 hours ago* (last edited 4 hours ago)

https://huggingface.co/RASMUS/Whisper_Finnish_finetuned_small_200k_samples

This could work if you look into setting up Whisper Search for other languages also

[-] flake@lemmy.dbzer0.com 1 points 3 hours ago

Thank youuu! I didn't know I could find models for other non-supported languages by default, that'll help me a lot!

[-] pe1uca@lemmy.pe1uca.dev 4 points 4 hours ago

Not sure what'd you consider lightweight, I've been using https://github.com/jhj0517/Whisper-WebUI with fast whisper.
The GPU integration has never worked well for me, but the CPU one works wonders.
You'll have to check if the models offer good results for those languages.

[-] flake@lemmy.dbzer0.com 3 points 3 hours ago

Thanks! I'm not sure what I should expect as lightweight when its about AI, i'd assume 2-4GB RAM max if possible and it seems with faster-whisper from this project according to the README it falls under 3-4GB RAM, so I guess that's fine!

There's no Finnish for Whisper last I checked but SGForce recommended a model for Finnish with it, so I guess that will do!

Thanks again for taking your time to help! :)