I recently ran into a frustrating problem: I needed high-quality data from YouTube for a side project involving LLMs. But extracting transcripts—especially in bulk—was either painful, slow, or hidden behind expensive subscriptions and API limits.
So I built: https://www.youtubetranscripts.io/
It lets you extract transcripts in bulk from any public YouTube channel, playlist, or video with a single click.
You can convert and download transcripts as .txt, .json, .csv, .srt, or .vtt, and customize what’s included—timestamps, video titles, channel names, etc.
I'll be constantly updating the app to improve extraction speeds. Currently, a channel with ~4,000 videos takes about 2 minutes to process.
Use cases for this tool: - Training data for LLMs - Content repurposing
Pricing model is pay-as-you-go. No subscriptions. Just purchase credits when you need them.
Would love to hear your thoughts or ideas for improving it.
Happy to answer questions about the tech stack, the scraping process, or decisions around pricing and UX.
Edmund