Uncrop a photo
AI image extender
Change aspect ratio
Portrait to landscape
Landscape to portrait
Instagram no crop
Full DP, no crop
Crop
Circle crop
Resize
Image splitter
Combine images
Blur faces
Rotate
Flip
Add text
Add a frame
Round corners
Remove background
Signature background
Adjust photo
Brightness
Contrast
Saturation
Black and white
Compress
JPG to PDF
Convert
GIF maker
Cut video
Compress video
MOV to MP4
Video to JPG
Audio converter
Video volume
Audio to video
AI image checker
Cropped photo checker
Remove EXIF data
QR code generator
Free video transcription, the video stays on your device
Video to text, a transcript with timestamps
Turn what is said in a video into text you can read, search and copy, with the time each line is spoken. A speech AI writes the words from a small sound-only copy of the video, and an AI cleans them up; the video itself never leaves your device. 3 minutes of speech a day is free, 10 minutes when you sign in, and typing or pasting a transcript yourself is always free.
- 3 minutes free a day, 10 minutes signed in
- The picture is never sent
- TXT, SRT, VTT and chapters
Your video stays on your device. Typing, pasting or opening captions, every style and saving the new video happen in your browser. Auto captions send a sound-only copy, then its words for a clean-up, and only when you press Make captions.
Make captions automatically
AITo write the words, your browser makes a sound-only copy of the video and sends it through uncrop.si to an outside speech AI service, which sends back the words and when each one is spoken. The words that come back (never the sound) then go through uncrop.si to an outside AI service for a clean-up of punctuation and filler words. The picture is never sent, and we keep neither the sound nor the words.
For the most accurate captions, choose the language spoken. Detect automatically works for most videos and for languages not in the list.
Transcribe only part of the video
Type a time like 1:05.5 or 65.5, drag the handles, or play the video and use the current position.
Spread evenly shares the video's length among the captions, longer ones getting more time. Tap along plays the video while you tap a button, or press Space, as each caption starts.
Times in the file are kept. A TXT file without times goes into the box above, ready to spread or tap along.
Play the video, then tap this button or press Space as each line starts. After the last line, tap once more when it ends. Each line runs until the next one starts.
Speech AI can mishear names, numbers and words over music or noise, and now and then it adds words nobody said. Check each line before you save. Lines it was unsure about are marked.
Parts without captions
AI tools for these captions
AITakes out filler words such as um and uh, and puts back punctuation and capital letters. The times stay as they are.
Every caption keeps its times. Undo brings your captions back.
A caption with hashtags from what is said, and a title for YouTube.
A short summary and chapters with start times for a YouTube description. Needs at least a minute of captions.
These tools send only the words and times of your captions, never the video or its sound, through uncrop.si to an outside AI service, when you press one of them; the clean-up also runs by itself right after auto captions, and that one is free. Each press is one of today's 5 free AI caption runs (15 when you sign in); very long captions can count as two or more.
Estimated timing. Typed and opened captions say when each line starts and ends, not each word, so each word's moment is estimated from its length. To move the words, change the line's start and end.
The text and box colors are close, so the captions may be hard to read. Text on a box reads best with a contrast of at least 4.5 to 1.
Saving waits until auto captions or the AI tool finish. Styles can still be changed meanwhile.
Video with the captions burned in
Keep this tab open and the screen on until the video is saved. Locking the phone, or switching to another app or tab, stops the save.
Keep this tab open until it finishes.
Transcript and caption files
SRT and VTT keep the times, for YouTube and other players that let viewers turn captions on and off. TXT is the words alone, one caption a line, or with the time each line starts. For YouTube, upload the SRT as well as the video, so viewers can turn captions on and off and screen readers get the words.
Your captions and style are kept in this browser only, so you can come back to them with the same video. Start again on the Captions tab clears them.
How to transcribe a video
Choose the video
Tap Choose a video and pick an MP4, MOV, WebM or MKV file, or drop it on the box. It opens on your device.
Pick the language
Leave Detect automatically, or choose the language spoken. Choosing it helps with short clips and with punctuation.
Make the captions
Press Make captions. Your browser makes a sound-only copy and sends it; the words come back with their times and are cleaned up by an AI.
Check and save
Read the lines marked Check this line, fix any word, then save the transcript as TXT, or as SRT or VTT with times.
Want the words on the video itself, or only the sound? Add subtitles to a video or Video to MP3.
What is sent, and what stays on your device
Your video is opened on your device and stays there. When you press Make captions, your browser makes a sound-only copy (one channel, compressed, without the file name or any tags) and sends it through uncrop.si to an outside speech AI service, which sends back the words and when each one is spoken. The words that come back (never the sound) then go through uncrop.si to an outside AI service for a clean-up of punctuation and filler words. The picture is never sent. We keep neither the sound nor the words; the speech service keeps nothing after it replies, and the clean-up service keeps requests only for a limited time for safety checks.
Opening the video, the transcript editor and saving the files all happen in your browser. A line under the tool counts what the page has sent since you added your video, so you can check it yourself.
Timestamps, SRT, VTT or plain text
Every line of the transcript keeps the time it is spoken, so you can jump to it, quote it or turn it straight into captions. Save it in the format the next app wants:
| File | What it holds | Use it for |
|---|---|---|
| SRT (.srt) | Numbered captions, each with a start and end time | YouTube, Facebook, video editors and most players. YouTube takes .srt and .vtt caption files (YouTube Help) |
| VTT (.vtt) | The web's caption format: the same captions with WebVTT timings | Web players and the HTML video element (W3C) |
| TXT (.txt) | The words as plain text, with or without times | Notes, articles, show notes, quotes and search |
Lines are kept short enough to read on screen: at most two lines of 42 characters, the length many subtitle style guides use, such as Netflix's for English (Netflix). For a plain transcript, save TXT without times.
A summary, chapters, a post or a translation
Once the transcript is there, the AI caption tools can work on its text: a short summary, chapters with start times you can paste into a YouTube description (YouTube needs at least three, the first at 0:00, each at least 10 seconds long, YouTube Help), a post with hashtags for YouTube, TikTok or Instagram, or a translation into another language. Each one is a free AI run: 5 a day, 15 when you sign in.
These tools send only the text of your captions, never the video or the sound. They are told to work only from what is said, but they can still get things wrong, so check what they write. The times of your captions never change.
Languages, accents and accuracy
The speech AI understands many languages and can tell which one is spoken. We tested it on read speech in English, Spanish, Portuguese, French, German, Arabic, Hindi and Japanese, and on English speech over noise and music. Speech AI can still mishear names, numbers and words over loud music, and now and then it writes words nobody said, so the tool marks the lines it was unsure about and keeps every line editable.
In some languages, such as Arabic, Hindi and Japanese, speech AI often leaves out punctuation. The clean-up after the transcript puts punctuation and capital letters back, and removes filler words like um and uh; the AI is told not to change what was said, so check any line that looks different.
Free minutes, and longer videos
Every day you get 3 minutes of auto transcription free, or 10 minutes when you sign in with Google, counted by the second, while the day's free pool for everyone lasts. A longer video is transcribed up to where today's free time runs out, ending at a pause so no word is cut, and the tool shows where it stopped. Typing, pasting or opening a transcript has no limit.
Making the sound copy and the transcript takes a few seconds for a short clip. The work happens in your browser, which needs WebCodecs to decode the sound (caniuse): Chrome and Edge, Firefox on a computer, Chrome and Samsung Internet on Android, and Safari 26 or later.
Questions
Is this video to text tool free?
Yes, within a daily allowance: 3 minutes of speech a day, or 10 minutes when you sign in, while the day's free pool for everyone lasts. Editing the transcript and saving TXT, SRT and VTT files are always free, with no watermark.
Is my video uploaded?
No. The video stays on your device. When you press Make captions, a sound-only copy is sent through uncrop.si to a speech AI service, then the words that come back go to an AI service for a clean-up. The picture is never sent. We keep neither the sound nor the words, and the speech service keeps nothing after it replies.
Can I get a transcript with timestamps?
Yes. Every line keeps the time it is spoken. Save SRT or VTT for times on every caption, or TXT with times for a readable transcript.
Which languages does it understand?
Many. It can detect the language or you can choose it. We tested English, Spanish, Portuguese, French, German, Arabic, Hindi and Japanese.
Can it summarize the video or make YouTube chapters?
Yes. After the transcript, press Summary and chapters: you get a short summary and chapters with start times, ready for a YouTube description. It counts as one of your free AI runs (5 a day, 15 signed in).
How accurate is it?
Clear speech comes out well; names, numbers, strong accents and speech over loud music are where speech AI slips. The tool marks lines it was unsure about, so check those before you use the text.
Can I transcribe only part of a long video?
Yes. Open Transcribe only part of the video and set the start and end. Only that part of the sound is sent.
Does it work on a phone?
It should run in phone browsers that can decode sound on the device: Chrome and Samsung Internet on Android, and Safari 26 or later on an iPhone (we have not tested phones ourselves yet). The video is read on the phone; only the small sound copy, and then its words, are sent.
Sources: YouTube Help, Supported subtitle and closed caption files, YouTube Help, Video chapters, W3C, WebVTT: The Web Video Text Tracks Format, Netflix Partner Help, English (USA) Timed Text Style Guide, caniuse, WebCodecs API
Open-source code and fonts this tool runs on your device: Mediabunny (MPL-2.0), unmodified, reads and writes the video and audio files; Noto fonts (OFL-1.1), Noto Sans Devanagari, Arabic, Thai and CJK and Noto Emoji, downloaded from uncrop.si only when your captions use a writing system your device has no font for; LAME (LGPL, source at SourceForge), the MP3 encoder (version 3.100) that makes the sound-only copy for auto captions in Safari and in browsers that cannot make an Opus one, built without its decoder for the browser by @mediabunny/mp3-encoder (MPL-2.0) and loaded only then.
Related tools
Add subtitles to a video Make captions automatically from the speech, or type, paste or open SRT and VTT files, then style them and burn them into your video. New Subtitle generator Make subtitles for a video automatically from its speech, fix any word, then save SRT or VTT or burn them in. No watermark. 
Video to MP3 Save the sound of an MP4, MOV, WebM or MKV video as an MP3, or as M4A with no quality loss. New 
Audio cutter Cut an MP3, M4A, WAV, OGG or FLAC file to the part you want, with fades, copied without re-encoding when it can be. New