Back
Question
Asked

Is there an API to extract the information from a YouTube URL?

I want to be able to extract information from a YouTube URL and pass it as context to an LLM. Are there any APIs to do so? If there aren't any, can anyone point me in the right direction to approach this?
Thanks in advance!


Can that help? nobinge.ai/?

Thank you! This might be useful, but their pricing is steep. I need to implement something myself.

There's a free tool to extract transcripts, which has a better UI and UX than others (no ads for instance): nobinge.ai/youtube-video-tran…

(I'm the maker. 😊)

Thanks! I need an api cause I need to bake this into my product

I can arrange that. Is the $29 plan too much for you?

Here's the link by the way: nobinge.ai/api

$29 is okay, but The 5 concurrent requests and 200 credits seem low :)

Yep. Actually, I didn't think about adding an endpoint to fetch a transcript.

So I just did it now. I'll make sure this new endpoint is unlimited for paying customers.

So if you want to start your trial, go ahead. You'll see the adjustment in your dashboard. 😊

I just tried the free tool and it’s super fast! 👏 great job!

If you are comfortable with Python or JS, I would call the YouTube Data API. They have native APIs for this, plus a bunch of other useful data. It's also very ChatGPT friendly.

They don't offer transcripts via their API unfortunately.

I manage some via their Captions API unless there is some nuance that I'm missing. We pay for transcription services or use Whisper and then upload them via the Captions API.

developers.google.com/youtube…

Well, I didn't know that and I spoke too fast!

I looked into it, but it doesn't seem like they have transcript available. They only have captions which I'm not so sure are accurate. Are the captions as accurate as in the case of uploading a audio extract to whisper?

The quality varies by the video. The only difference between the two is that captions include timestamps with the transcriptions.

I'm glad you found a working solution. Check that API out if the pypi project ever goes out on you. It should be quick to get it back up and running with it.

Hey y'all! Thanks for the super helpful replies. I was able to get a fast api tiny service up and running that extracts the transcript using this python library: pypi.org/project/youtube-tran…

It's super fast!!!