Is there a YouTube summarizer API?

TranscriptFetch gives you the transcript, which is the hard part: YouTube blocks most servers that fetch captions. Pass it to the model you already use, with your own prompt, and you have a summarizer that works on TikTok and Instagram too.

Why not let the model summarize the link directly?

Most models cannot open a video link, and the ones that try often get blocked. Fetching the transcript first gives the model the full text every time.

What about long videos and podcasts?

Long videos with captions come back in one call. Without captions, AI transcription runs in the background: poll the job or pass callback_url and the finished transcript is posted to you.

Can I show where each point came from?

Yes. Every line has a timestamp, and the Markdown version links each paragraph to its moment in the video.

All use cases

Build AI video features: summaries, notes and Q&A

Every video feature built on a language model starts the same way: get the words. TranscriptFetch returns the transcript of any YouTube, TikTok or Instagram link your users paste in, from captions when the video has them and by AI transcription when it does not. Your model turns it into summaries, notes, chapters, quizzes or answers.

Create a free account Read the docs

How it works

01
Your user pastes a link
A YouTube video or Short, a TikTok, an Instagram Reel, or a direct audio or video file.
02
Fetch the transcript
One call returns the text with timestamps, plus Markdown with clickable timestamps if you want it ready to drop into a prompt.
03
Your model does the rest
Summarize it, write chapters from the timestamps, pull quotes, make flashcards, translate it, or answer questions about it. You choose the model and keep the prompt.

Example

Node: a summary of any video link

javascript

const r = await fetch("https://transcriptfetch.com/api/v2/transcripts/video", {
  method: "POST",
  headers: { Authorization: "Bearer tf_live_YOUR_KEY", "Content-Type": "application/json" },
  body: JSON.stringify({ video: userLink, markdown: true }),
});
const { data } = await r.json();

// data.markdown: title, channel and paragraphs that open with linked timestamps
const summary = await llm(`Summarize this video in five bullet points:\n\n${data.markdown}`);

What you get

One response shape for YouTube, TikTok, Instagram and file links, so one code path covers them all
Title, channel, duration, language and thumbnail with every transcript
AI transcription when a video has no captions, with a job to poll or a webhook for long videos
Failures are free, so a dead link costs you nothing

Go deeper

Questions

Is there a YouTube summarizer API?
TranscriptFetch gives you the transcript, which is the hard part: YouTube blocks most servers that fetch captions. Pass it to the model you already use, with your own prompt, and you have a summarizer that works on TikTok and Instagram too.
Why not let the model summarize the link directly?
Most models cannot open a video link, and the ones that try often get blocked. Fetching the transcript first gives the model the full text every time.
What about long videos and podcasts?
Long videos with captions come back in one call. Without captions, AI transcription runs in the background: poll the job or pass callback_url and the finished transcript is posted to you.
Can I show where each point came from?
Yes. Every line has a timestamp, and the Markdown version links each paragraph to its moment in the video.

Get a free API key

Every account gets 50 free credits a month, no card needed. Failed requests are free.