Keep the file you already have
Export a voice memo or keep the meeting M4A. Do not upload it to a cloud notetaker first if the point is that it stays here.
Drop an MP3, M4A, WAV or MP4 you already have and get a timestamped transcript in this browser.
Runs in this browser. The file never leaves this device.
Audio to text here is a file you already have: MP3, WAV, M4A or MP4, up to 15 minutes, in this tab. Nothing is uploaded. A voice memo, or a WhatsApp message saved as M4A or MP3, belongs on Voice memo to text. A WhatsApp .opus or .ogg file is refused. A Zoom file from Documents/Zoom belongs on Transcribe a Zoom recording.
Convert an MP3 to text in this browser. WAV and MP3 are fine. MP4 works when your browser can decode the audio track. A file extension is not a guarantee: if decode fails, export a standard WAV or MP3 and try that copy.
Free transcription here stops at 15 minutes. Browser Whisper holds the recording in memory. Fifteen minutes is the trial limit, not a claim about accuracy. A two-hour meeting belongs in Vemoir for Mac, where local models are built for that length. Splitting a long file into parts still spends a weekly run per successful pass.
Speech to text for a full meeting is Vemoir for Mac: it captures microphone and system audio during the call, keeps a library, and uses stronger on-device models. The browser does not record the meeting and does not join as a bot.
This page does not run a speaker-separation model. If the file already contains names or “Speaker 1” labels, those are kept. Who-said-what on a raw two-person tape is a Mac job.
Vanilla Whisper often writes Standard German from Swiss German speech. There is no dialect model in this browser download. Serious Swiss German work stays in Vemoir for Mac.
Export a voice memo or keep the meeting M4A. Do not upload it to a cloud notetaker first if the point is that it stays here.
MP3, M4A, WAV or MP4, up to 15 minutes. The page checks type, size and length before it downloads the model.
Choose the language spoken in the recording. The tool does not translate.
The first visit downloads Whisper into this browser. Keep the tab open.
Copy the text, or download TXT, SRT or VTT. Nothing is sent to Vemoir.
No. After the speech model is cached, you can open the Network tab and confirm that the audio is not posted anywhere. Inference stays in this tab.
Browser Whisper is a trial. Longer meetings belong in Vemoir for Mac, where local models are built for that length.
Drop an MP3, M4A, WAV or MP4, up to 15 minutes. There is no account, and nothing is uploaded.
No. There is no signup. Three successful runs a week are stored as a count in this browser.
No. You drop a file you already have. Live capture is Vemoir for Mac.
Not on a raw tape. Names already in the file are kept. Speaker separation is a Mac job.
Vanilla Whisper often writes Standard German. There is no dialect model in this browser download.
The card stays on the page. You can still try the sample. New files wait until next Monday, or you continue in Vemoir for Mac.
Vemoir for Mac captures the call on the machine, then writes minutes, decisions and actions locally.