100% browser-based · zero uploads

YouTube Video Summarizer

Summarize any YouTube video in seconds — paste the transcript, pick a length, and get a concise AI summary. Free, no account required, and nothing leaves your device. The AI model runs entirely in your browser.

Supports youtube.com/watch, youtu.be, youtube.com/shorts, and youtube.com/embed links.

How It Works

1

Enter the YouTube URL

Paste any YouTube link. The tool validates the URL and extracts the video ID — no data is fetched from YouTube.

2

Paste the transcript

Use YouTube's built-in transcript panel to copy the video text, then paste it in. Nothing is uploaded.

3

AI summarizes locally

The AI model runs right here in your browser — not on a server. Long transcripts are chunked automatically.

4

Copy or download

Copy the summary to your clipboard in one click, or save it as a .txt file.

How to Summarize a YouTube Video Online with AI

  1. Enter the YouTube video URL and click Continue.
  2. On YouTube, click (More actions) below the video, select Show transcript, click Toggle timestamps to hide time codes, then copy all the text.
  3. Paste the transcript into the text area.
  4. Select a summary length: Short for a quick overview, Medium for the key points, or Long for a detailed condensed version.
  5. Click Summarize. On first use, a one-time model download (~343 MB) is required — your browser saves it locally for instant reuse.
  6. The summary appears in the output area. For transcripts over ~700 words the tool splits the input into chunks automatically and produces a final combined summary.
  7. Click Copy to copy to your clipboard, or Download .txt to save the summary as a file.

When to Use Each Summary Length

Length Output Best for
Short 3–4 sentences Quick overview, social media captions, TL;DR
Medium 5–7 sentences Meetings, lecture notes, podcast summaries
Long 8–10 sentences Long-form documentaries, tutorials, interviews

Features

  • Your transcript stays private: The AI model runs entirely in your browser — nothing is uploaded, stored, or shared.
  • No account required: Open the page and start summarizing straight away.
  • Three summary lengths: Short, Medium, and Long to match your needs.
  • Automatic chunking: Transcripts over ~700 words are split into segments, each summarized, then combined into a final summary.
  • Works offline: After the one-time model download, the tool runs with no internet connection.
  • Flexible output: Copy text to your clipboard or download it as a plain .txt file.
  • Powered by DistilBART: A fast BART-derived model trained on CNN/DailyMail — strong factual extractive-abstractive summaries. One-time download of ~343 MB.

Frequently Asked Questions

Is my transcript uploaded to your servers?

No. Your transcript never leaves your device. The entire process runs inside your browser — nothing is sent to or stored on any server.

Why do I need to paste the transcript myself?

Browser security (CORS) prevents websites from fetching YouTube data directly. Pasting the transcript yourself also means the tool works with any captioned video — with no API key, no backend, and no data leaving your browser.

How do I get the YouTube transcript?

Open the video on YouTube, click (More actions) below the player, select Show transcript, click Toggle timestamps to hide the time codes, then select all the text and paste it here.

Is this free to use?

Yes, completely free. No account required, no usage limits, and no hidden charges.

Why is the first summarization slow?

The first time you use the tool, your browser downloads the AI model (~343 MB — encoder and decoder ONNX files). After that one-time download, the model loads instantly from local storage and the tool works offline too.

Does it work offline?

Yes. After the model is downloaded on your first use, it is saved to your browser's local storage. You can then summarize transcripts with no internet connection.

Which AI model does this use?

DistilBART (distilbart-cnn-12-6) by Facebook/Hugging Face — a distilled version of BART trained on CNN/DailyMail news. It runs as a quantized ONNX encoder + decoder via Transformers.js in your browser using WASM or WebGPU. The one-time download is ~343 MB.