Private by design
Audio to text, and your audio never leaves your device.
Free transcription with speaker labels, running entirely in your browser. No signup, no subscription, and nothing uploaded. Built for confidential work.
Takes a little extra time, and your audio still never leaves your device. Labels go by how voices sound, so treat them as a best effort: one person can be split across two or three labels, similar voices can be merged, and someone joining partway through may be folded into a speaker already talking. Giving a number usually helps. Check them before relying on them.
Specifying this number improves accuracy. Count people who really take part, whenever they join. Someone who only says a word or two is better left out.
Timestamps each word instead of each phrase, so short interjections like "Yeah" go to the right person. On a busy four-person recording this corrected about 1 word in 5; it changes little when people talk in long turns. Adds roughly 40% to the time on a computer, and almost nothing on a phone.
Phone audio loses the high frequencies that separate consonants, so they are put back before transcribing. Only phone-quality recordings are treated, and you are told when it happens. Also even out volume is for one case: someone far from the microphone and much quieter throughout. It cost accuracy on other recordings, so use it only if a quiet person came out badly.
Worth filling in. The speech model knows ordinary English, not your subject, so a word it has never heard becomes whatever sounds closest, and it usually gets that word wrong every single time it comes up. A sentence is enough, and the names and jargon in it are picked out and watched for. It cannot guess a word you do not mention, so name the people and the terms that matter.
The first time you use it, it takes a minute or two to set up. After that setup is quick. Works with MP3, WAV, M4A, OGG, FLAC and AMR. Long recordings are handled in parts, so most files are fine whatever their length.
How is this private?
Most transcription services upload your recording to their servers to process it. MeetingScribe does not. When you press Transcribe, the speech-recognition model is downloaded into your browser once, and then your audio is processed right here on your own device.
Your audio file, the transcript, and the meeting-notes prompt all stay on your computer. There is no account, no upload, and no copy of your recording sitting on someone else's server. There is no paid tier and no server that processes your content: the whole tool is the page in front of you. You can verify it: open your browser developer tools, watch the Network tab, and you will see the model download but never your audio going out.
If you paste the notes prompt into an outside AI chat to format your notes, that is your own action on a service you choose, and only the text you paste goes there. This is designed for confidential workflows where uploading a recording is not an option. We do not claim any specific regulatory certification; you remain responsible for your own compliance obligations.
On your device
Transcription and speaker separation run locally in your browser.
No upload
Your audio is never sent to us or anyone else.
Yours to keep
Download the text, subtitles, and notes prompt. Nothing is stored by us.
Made for confidential work
If you cannot or will not upload a recording to a cloud service, this is for you.
Lawyers
Client interviews and privileged calls that cannot leave your control.
Learn moreTherapists
Session recordings that must stay confidential.
Learn moreJournalists
Source interviews you need to protect.
Learn moreEveryone else
HR, doctors, researchers, and anyone handling sensitive audio.
Learn moreNew here? See what you can and cannot do.
What you get
One transcription produces everything below, and all of it stays on your machine.
A transcript you can actually edit
The text appears in an editor in the page, so you can fix misheard names and technical terms before you download anything. Speech models mishear proper nouns more than ordinary words, and correcting them once here beats re-running the whole file.
Transcription with speaker labels
A second model separates voices and labels each turn, and you can rename the labels to the people who were in the room. It works best with one to three voices, and it does not identify anyone: it only tells voices apart inside the one recording. More on how speaker labels work.
Subtitles with timestamps
Every run also exports SRT and VTT files, timed to your audio, for a video editor, YouTube, a desktop player, or a web page. See the guide to SRT and VTT for which format each tool expects.
Meeting notes, on your terms
One click copies a ready-made prompt containing your transcript and clear instructions, so any AI chat can turn it into a summary with action items. We deliberately did not build a one-click cloud summary, because it would send your transcript to a server and quietly undo the point of the tool. That reasoning is on the about page.
Common questions
- Is it really free, and what is the catch?
- Yes. This is free transcription without a subscription: no account, no file limit, no trial that expires, and no paid tier at all. The site is funded by advertising, which only loads if you accept cookies. The honest catch is what it can do, not what it costs: the first run takes a minute or two to set up, very long recordings in compressed formats can be more than a browser tab will hold (saving as WAV avoids that), and on really difficult audio a big paid service will still beat it.
- How can transcription work without uploading my file?
- The speech-recognition model is downloaded into your browser once and then runs on your own processor or graphics card. The model travels to your audio instead of your audio travelling to the model. After that first download it is cached, and transcription works even with your internet disconnected.
- What file formats can I use?
- MP3, WAV, M4A, OGG, FLAC, and AMR. AMR files from phone recorders, which no browser can decode natively, work here because the tool loads its own AMR decoder and runs it in your browser. For video, export the audio track and transcribe that.
- Can it tell me who said what?
- Yes. Speaker separation runs on your device alongside the transcription and labels each turn, and you can rename the labels to real names. It works best with one to three voices, and it identifies nobody: it only tells voices apart within the one recording.
- Can I get subtitles?
- Yes. Every transcription also produces SRT and VTT subtitle files with timestamps, ready for a video editor, a player, or an HTML5 video element.
- How do I get meeting notes or a summary?
- After transcribing, one click copies a ready-made prompt containing your transcript plus clear instructions to your clipboard. Paste it into whichever AI chat you already use and you get a summary, key points, and action items. Nothing is sent by the copy itself, and there is deliberately no built-in cloud summary, because that would mean sending your transcript to a server.
For the full pipeline, every network request the page makes, and two ways to verify the privacy claim yourself, read how it works. If something is wrong or missing, tell us.
Guides worth reading first
Practical writing on the questions that come up before and after a transcription.