Simple Website with subtitles generation and audio to text conversion

[Kabanosk]

Hi! Today I heard about this super open-source project. I want to contribute by making simple web application, which will be GUI for this model. So I spent couple hours and I would be glad if someone help me improve it. Repo with the website is on my github profile.

Edit#1: You can download subtitles in vtt and srt file type.
Edit#2: I added hosting with Docker.

[zwoefler]

Will take a look when I have time. Thanks for doing this!

[Kabanosk]

Thanks for your appreciation ;)

[jmichelin]

ooh looks fun! I will jump in too

[Adriankn0rr]

How have you managed to host the whisper model. I‘ve tried AWS lambda layer, but the model seems to large for AWS. Would be very thankful to figure out a solution to host my Python code along the whisper model.

[Kabanosk]

I host it on the localhost, so I don't have that problem. I think that it is possible to download the models and run the inference without downloading model every time. Do you think that will help?

[jonathanjfshaw]

You can run whisper itself in the browser: https://github.com/ggerganov/whisper.cpp/tree/master/examples/whisper.wasm

[dankentfield]

@Adriankn0rr have you tried using a lambda layer for the model?

[baibhavKumar1]

But it's in C language how to work with that?

[jonathanjfshaw]

I find this huggng face space to be very well done: https://huggingface.co/spaces/aadnk/whisper-webui

[AlanaRm-rf-me]

Thank you for this.

More the reason to head back into web dev, and seeing how ya have this working!

[gsweats]

Lol, lead FS web dev and DevOps person. Thinking about moving into this from web. Haha, I'll have to set a reminder to follow you and check how that goes, I think once AI has you, you are its no matter what. There's no going back.

[Kabanosk]

Thanks a lot for your kind words, @GregSweats :D

[Cafe-Soluble]

Worked well ! Thank you :)

[Kabanosk]

Thank you very much! :D