Show HN: I made a website that converts YT videos into step-by-step guides

This is a brilliant and useful application of LLM technology, I'm impressed.

One question- On the backend, is it downloading each video CC (closed-caption) transcript and feeding that into a tuned prompt? What happens for videos where this is missing? Asking because I've noticed CC is occasionally unavailable for some YouTube videos.

If you cared to have a fallback, a potentially interesting experiment / solution for such cases is to download the video, extract the audio to a WAV file, then through the audio through Whisper [1] to generate the transcript. Using CPUa, it will still be incredibly intensive and slow, generally not much faster than real-time (e.g. a 5 minute clip will take on the order of ~5 minutes to complete transcription). However, with Whisper running on a fancy GPU it is insanely faster, between 100-200x faster, meaning even for long videos, generating the transcripts will complete in only a few seconds.

Great job @aka_sh!

[1] https://github.com/openai/whisper

p.s. Is there any chance you'd open source your code? Or do you plan to turn this into a business? The code itself is exactly a huge moat, and it'd be cool to see how you did this. Cheers.

p.p.s. stepify.tech app is currently crashing out to a heroku error page when I try to submit a YT link.

I made something a little similar, but just as a little cli script that I run locally for myself. You can input a url for a YouTube video, podcast link or local audio/video file. It transcribes it with whisper and outputs the full transcript in one text file and I use another model to summarize it into a bullet list in a separate file.

I so appreciate these open source/access models allowing us to build these kinds of tools without having to pay and send our data to openai.

Great work. A few ideas

1) Speed : the site is often showing heroku errors. Seems like you are running the entire processing in the request-response cycle. If not already done, please try to use a queueing system to perform async processing - and then let the user know when their video is ready to view as steps (probably via email or browser notifications). This will stop your site from crashing frequently and you'll be able to scale to many users very quickly.

2) Please add link-backs to the specific time in the video from where the step is shown.

Cheers!

For the "Paid" or "Pro" version, let me have a browser extension that replaces ALL OF YOUTUBE with your text based breakdowns.

// I'm not really kidding! Because boy do I hate 15 minute videos with the one CLI command you need buried like a needle in a haystack. Seeing the nonsense distilled into a handful of straightforward steps is so refreshing. Awesome work!