How to convert a video to text
- Drop your video. MP4, MOV, WebM, MKV, or audio such as MP3, WAV and M4A, up to 10 minutes. Or press Try a sample clip.
- Check the settings. Leave the language on automatic or pick it, keep Transcript selected, and press Transcribe to text. The tool picks the best speech model your device can run and says which one.
- Read and copy. The text appears as paragraphs. Turn timestamps on or off, click a timestamp to hear that part, and press Copy all.
- Download. TXT for notes and blog drafts, DOCX for Word or Google Docs, SRT or VTT if you also want subtitles.
What you can use the text for
- Notes from lectures and meetings: search the text instead of scrubbing through the video.
- Blog posts and newsletters: start from what you already said on camera.
- YouTube descriptions and chapters: the timestamps show where each topic starts.
- Quotes for articles: copy the exact words, with the time to check them.
- Subtitles: switch to Subtitles at any time for a timed, editable caption track.
Accurate enough to use
The default model on a computer with WebGPU is Whisper large-v3 turbo, the most accurate open speech model that runs in a browser. On our real-speech tests it got 2% of words wrong in Spanish and 4.5% of characters wrong in Arabic. Phones and older computers get a smaller model, and the tool tells you which. Names, numbers and technical words are where any speech model slips, so read the text before you publish it.
Private by design
Most video to text sites upload your file to their servers. This one doesn't: the speech model downloads to your browser once (the size is shown before it starts) and the video is transcribed on your own computer. That makes it safe for interviews, client calls and anything you'd rather not share.
Languages
Automatic detection covers about 99 languages. Pages with our measured accuracy: Spanish, Arabic, Urdu.
Questions
Is this video to text converter free?
Yes, with no sign-up and no watermark. Files up to 10 minutes are free. It costs us nothing to run because your device does the work.
Can I get the text with timestamps?
Yes. Timestamps are on by default, one at the start of each paragraph, like [1:05]. Untick Timestamps for plain text. SRT and VTT always carry the full timing.
Can I download the transcript as a Word document?
Yes. DOCX opens in Microsoft Word, Google Docs, Pages and LibreOffice. Urdu and Arabic paragraphs are set right to left.
Does it work with Hindi, Urdu and other languages?
It understands about 99 languages. We measured Spanish, Arabic, Hindi and Urdu on real speech; see the language pages for the numbers. For Hindi and Urdu, pick the language by hand: automatic detection can confuse the two.
Is my video uploaded?
No. It is decoded and transcribed in this browser tab. Only the speech model is downloaded, from Hugging Face's CDN, and your browser keeps it for next time.
Can it transcribe a YouTube link?
Not from a link. Download the video you have the rights to, then drop the file here.