AI video summarizer
AI video summarizer: key points from a YouTube link or your own file
An AI video summarizer reads what is said in a video and writes it back in a few lines: here you paste a YouTube link or upload a recording and get 5 to 7 key points, detailed notes, an article or answers in a chat, free on any YouTube video with captions. Files and uncaptioned videos are transcribed first, from $0.033 per minute. The AI works from speech, not from the picture.
Updated
On this page
How the AI video summarizer works
Summarizing a video takes four steps, and the AI starts only at the one where you press a button.
- Paste a link or drop a file. Any YouTube address works: a watch page, a youtu.be short link, Shorts or an embed. A file can be MP4, MOV, MKV, WEBM, MP3, M4A, WAV or another common format.
- Wait for the transcript. A captioned YouTube video opens with its caption text in a few seconds. A file, or a video with no captions, goes through Whisper speech recognition first, and a long one reports its progress in minutes.
- Pick an action. Sixteen tiles sit next to the transcript in four groups: Understand, Create, Learn and Work with text. Nine of them run today, and nothing is generated until you press one.
- Read, copy or save the result. The output appears beside the player, ready to copy, and downloads as PDF or DOCX.
Results come in the language of the video; to read a Spanish interview in English, switch the transcript to English first. Long recordings are not cut off after the opening minutes either. The transcript is condensed part by part, so a three-hour lecture is summarized from its first sentence to its last.
Nine actions in the AI video summarizer
Nine AI actions work on any transcript today, and each turns the same words into a different kind of text.
| Action | What comes back | Typical use |
|---|---|---|
| Summary | 5–7 key points, one sentence each | deciding whether a talk is worth an hour |
| Detailed notes | the whole recording split into topics, with facts, figures and names under each | studying and revision |
| Article | a blog post with a headline, 3–5 sections and a short conclusion | reusing a webinar as written content |
| Social post | three drafts: one for X, one for LinkedIn, one for Instagram | announcing an episode |
| Custom prompt | an answer to your own instruction, drawn from the transcript only | anything the other tiles don’t cover |
| Chat with video | replies to questions about what was said | finding one detail without rewatching |
| Pro Transcript | the full text with punctuation and paragraphs, minus filler words and false starts | quoting or publishing a talk |
| Translate text | the transcript in another language | reading a foreign clip |
| Export as file | TXT, Markdown, SRT or VTT | notes apps, subtitles, archives |
Detailed notes have their own guide on turning a video into structured notes. Two kinds of recording get pages too: summarizing a recorded lecture covers class recordings and long talks, and the audio summarizer for podcasts and voice memos covers files with no picture at all. How these actions compare with a quota-based tool is set out in MediaScribe next to NoteGPT.
Translation is the one action with a different price on the site: AI translation uses minutes, while the Chrome extension translates captions for free on the YouTube page itself, as the guide to translating a YouTube video explains. Exported subtitle files can be reshaped later with the subtitle converters, which run in your browser.
Seven more tiles carry a “soon” label: key moments, chapters and timecodes, slides, questions and answers, a quiz, flashcards and a mind map. They don’t run yet.
A YouTube link or your own file
The source sets the price: a video that already has captions hands the AI ready text at no cost, while a file or an uncaptioned clip needs speech recognition first.
- YouTube with captions, uploaded by the author or generated by YouTube: the text comes from the caption track, and every AI action is free with no sign-up. If plain words are all you need, YouTube to text gives them without any AI. The YouTube transcript itself stays free for good.
- YouTube without captions: Whisper transcribes the sound for one minute of balance per minute of footage.
- Your own file: up to 100 MB without an account, up to 500 MB and 4 hours after sign-in. The guides to transcribing a video file and to turning an audio recording into text list every accepted format.
YouTube generates captions automatically for speech in many languages, so the free route is the common one. Links from Vimeo, TikTok, Google Drive or Dropbox are not accepted: download the clip, then upload it here.
What a video summary looks like
A summary is a bulleted list of 5 to 7 sentences, each carrying one point, with no introduction and no filler.
Illustrative output for a 25-minute talk on job interviews, not a real recording:
- Read the company’s news from the last two years, not just its home page.
- Prepare three stories from past work, each with a problem, your action and a measured result.
- Answer salary questions with a range based on published pay data for the role.
- Ask the interviewer how success in the first 90 days will be judged.
- Send a short follow-up within 24 hours that repeats one point from the conversation.That is what “summarize the key points of the video” means here: the points, in the order the speaker made them. Detailed notes on the same talk run several times longer: a topic line for each part, then the advice, numbers and names beneath it. A summary helps you decide whether to watch; notes are what you keep when the content matters later.
Who uses a video summarizer, and for what
A video summarizer pays off wherever the words in a recording matter more than the picture: lectures, interviews, webinars, tutorials and podcasts.
- Students summarize a video lecture before the seminar and keep detailed notes for the exam; a class recorded in Zoom is handled like a YouTube link.
- Researchers and journalists check whether a two-hour interview mentions their subject at all, then quote the exact line from the transcript with timestamps.
- Creators turn their own upload into an article and three social drafts in the time it takes to read them.
- Teams send colleagues the key points of a recorded webinar or all-hands meeting instead of a 60-minute link.
- Viewers summarize this video or that one before giving it an evening, from a conference keynote to the text of a TED talk.
All five start the same way: summarize a video once, then decide what deserves full attention.
Your own prompt instead of copying into ChatGPT
A custom prompt does the job of a “YouTube summary with ChatGPT” routine without the copy and paste: you type an instruction, and the answer is drawn from the transcript alone.
- List every book and author the speaker mentions.
- Write five revision questions with short answers.
- Summarize only the part about pricing.
- Turn the setup instructions into a numbered checklist.
When the transcript doesn’t hold the answer, the reply says so instead of inventing one. Chat with video works on the same rule, one question at a time, which suits a two-hour podcast where you need a single figure.
What’s free and what uses minutes
AI analysis of a captioned YouTube video costs nothing; minutes are spent only on speech recognition, from $0.033 per minute of recording.
| What | Cost | Conditions |
|---|---|---|
| Transcript of a captioned video | free for good | no sign-up, no limits |
| AI actions on a captioned video | free | no sign-up; on a day of very high demand, free AI may pause until the next day |
| First file or uncaptioned video | free | no account; in full up to 10 minutes, the first 10 minutes of a longer one |
| Speech recognition after that | 1 minute of balance per minute of recording | from $0.033 per minute; the price is shown before you start; every AI action on that recording included |
| AI translation on the website | minutes | free in the Chrome extension |
| Sign-up gift | 10 minutes of transcription | once per account, used first, expire after 7 days |
| Packs | 1 h for $2.99 · 3 h for $7.99 · 10 h for $19.99 · 30 h for $58.99 | one-time payment, no subscription; refunds within 14 days |
A 45-minute webinar file, for example, uses 45 minutes of balance once; a summary, notes, an article and a chat about it add nothing to that.
What the AI video summarizer doesn’t do
The summarizer works from speech only, so anything shown on screen but never said, such as slide text, code or a chart, stays out of the result.
- No timestamps inside a summary or notes; the transcript beside them keeps every
[m:ss]mark. - No links other than YouTube: Vimeo, TikTok and cloud drives need a downloaded file.
- No live summaries of a meeting or call in progress; upload the recording when it ends.
- No speaker names: the transcript is one stream of lines, whoever talks.
- Nothing to work with in a silent clip or one with music only.
- The seven “soon” tiles don’t run yet.
What you can rely on instead: the price appears before anything starts, and minutes are charged only after a transcription succeeds.
Frequently asked questions
Is the AI video summarizer free?
Yes, for YouTube videos with captions: every AI action on them is free, with no sign-up. Files and videos without captions need speech recognition, from $0.033 per minute, after a free first recording of up to 10 minutes.
Do I need an account?
No, not for a captioned YouTube video and not for your first recording. Yes for whole files after that: sign-in is a link sent to your email, with no password to create.
Can it summarize a YouTube video without captions?
Yes, once the sound is transcribed. Whisper turns the audio into text for one minute of balance per minute of video, and every AI action on that video is included afterwards.
Where does my uploaded video go?
The original file is deleted from our server as soon as the sound is extracted. The sound goes to OpenAI’s Whisper, and its pieces are deleted once the transcript is ready; the transcript then goes to an OpenAI model to write the result you asked for.
In which language does the summary come back?
In the language of the video. For another language, pick it in the language menu first: the author’s own translations are free, other languages go through AI translation paid with minutes from your balance.
Can it read slides or text shown on screen?
No. The summarizer works from speech, so slide text, code or a chart that nobody reads aloud never reaches the result.
What if free AI is unavailable?
On a day of very high demand, free AI on captioned videos may pause until the next day, and the page says so. The transcript itself stays available.