Available captions, useful text
Read an available caption track in its original order, find the passages you need and save the text in a useful format.
Opening a transcript sends your text to our server for checking, without persistent storage. AI processing starts only when you press its creation button.
A small example
This is a demonstration. A real document preserves the times and words of its caption track.
00:00 Today we will explore how models learn.
00:08 We will start with data and examples.
TXT, SRT and VTT files up to 1 MiB; pasted text up to 100,000 characters. Plain text and start-only lines export to TXT/MD. SRT/VTT require a real start and end for every cue.
AI works with the full transcript. Text over the limit is not sent.
Download important work to keep a copy if browser data is cleared. In a rolling 24-hour window: up to 25 successfully imported YouTube tracks and 10 successful AI results. Per rolling 30 days: up to 100 YouTube tracks and 50 AI results. Reading and export do not spend your allowance.
Read about data and limitsNo. An accessible existing caption track is required. If there is none, you can upload or paste your own transcript.
No. The readable view groups words into paragraphs; AI versions stay separate and remain linked to their source track.
These pages and forms remain available. Submission shows a temporary readable server result; the local library, search and AI panel require JavaScript.