Free, no sign-up, no upload
Transcribe audio to text
Add a recording or a video and get its words back as paragraphs you can search, copy, or save as a text file. The transcription runs in your browser, so the file stays on your device.
Read on your device. Never uploaded.
Transcript
How to use it
From a recording to a transcript in three steps
1. Add a recording
Drop an audio or video file on the box above, or click it to choose one. Your browser reads the file on your device.
2. Pick the language and model
Leave the language on automatic detection or choose it, keep the smaller model or take the larger one, and tick the box if you want timestamps.
3. Search, copy, or save
Paragraphs appear as each part of the file is done, and you can search them at once. When the run ends, copy the text or download it as a TXT file.
A real result
What the transcript looks like
A short test recording, the same sentence three times with a pause between, transcribed by this page with timestamps on and "place" in the search box. It is shown as the page wrote it, mistakes included.
What you get
A transcript to read, search, and keep
Paragraphs you can read
A new paragraph starts at a pause of 2 seconds or more and, in a long stretch of talk, at a sentence end, so the text reads in blocks instead of one long line.
Timestamps if you want them
Tick the box and each paragraph starts with its time, such as [00:04:12]. The time is approximate, one per paragraph, not one per word.
Search inside the transcript
Type a word or phrase to highlight every match, see how many there are, and step through them with the arrows or the Enter key.
Copy or download TXT
Copy text puts the whole transcript on your clipboard. Download TXT saves the same text as a plain file named after your recording.
Video files too
Add a video and the page transcribes its sound track. Nothing plays, and the picture is not decoded.
Long recordings in parts
Where your browser can, the file is read a slice at a time; either way it is transcribed in windows of about three minutes, so you can read the start while the rest is still running.
Formats, models, and limits
Files. Any audio or video file your browser can decode. Our automated test transcribes a WAV file, an MP4 video, and a WebM file in Chromium; other formats, such as MP3 or M4A, depend on the decoders in your browser, and the page tells you when it cannot read one. Where a browser has no decoder that reads a file in slices, the page decodes the whole file at once instead, and then refuses files over 500 MB or longer than 1 hour.
Models. The smaller model is the default and downloads about 60 MB on your first file. The larger one downloads about 265 MB and needs more memory. How the two compare in accuracy and speed is not published yet: we measure both on this engine first. What the page downloads and what your device needs explains the files and the parts they come in.
Time. Your device does the work, so the time depends on its processor, on the model, and on the length of the recording. The page shows an estimate of the time left, measured on your device while it runs, and Cancel stops a run at any point.
What it does not do. It does not label speakers, translate, fetch a recording from a link, or take dictation from a microphone. The about page explains each one.
Privacy by design
Your recording stays on your device
No upload
The page reads your file with your browser's File API, and the site has no upload endpoint. Its security policy lets the page's scripts connect only to this site.
No account
Nothing to sign up for and nothing to install. The pages set no cookies and run no analytics.
Tested on every change
An automated browser test transcribes sample files and fails if any request leaves this site or carries data.
FAQ
Common questions
Is it free?
Yes. There is no account, no paid plan, and no count of files per day. What limits a file is your device: it does the work, so a long recording takes a while. In a browser that cannot stream the file's sound, the page reads at most 500 MB and 1 hour per file.
Is my recording uploaded?
No. Your browser reads the file and the transcription runs in your browser tab. The only downloads are the page, the transcription engine, and the model files, all from this site, and nothing goes back. You can check that nothing is uploaded in your browser's developer tools.
Does it tell speakers apart?
No. There are no speaker labels, so an interview or a meeting comes out as one text, and a new paragraph marks a pause, not a new speaker. Add names yourself after you download the TXT file. What the tool does not do lists the other gaps.
Does it translate?
No. The transcript is in the language spoken in the recording. Choose that language, or let the page detect it.
How long does an hour of audio take?
It depends on your device and the model you pick, so the page does not promise a speed. While it runs, it measures how fast your device is going and shows an estimate of the time left.
Which formats work?
Our automated test transcribes a WAV file, an MP4 video, and a WebM file with Opus sound in Chromium, the engine behind Chrome and Edge. Other audio and video files work when your browser can decode their sound; if it cannot, the page says so.
Are the timestamps exact?
No, they are approximate: one time at the start of each paragraph, rounded down to the second. They are good for finding a passage in the recording again, not for subtitles.
Does it work offline?
No. The page needs a connection to load, and on your first file it downloads the engine and the model. Your browser keeps the model afterwards, so later files skip that download.
Curious what happens to your recording?
It stays in your browser tab. See what the page downloads, what your browser keeps, and how to check it.