Audio to text, free and private
Drop a recording and read the transcript. The file never leaves your computer, and there is no sign-up or daily limit.
Drop an audio or video file
MP3, WAV, M4A, MP4, MOV, WebM
First use downloads about 80 MB, once.
How it works
- 1Add the audio
Drop a file or press Record. Video works too, the tool reads its sound.
- 2Let the model load
The first time, your browser downloads the speech model, about 80 MB. Next time it starts at once.
- 3Copy or download
The text appears as it goes. Save it as plain text, or as SRT and VTT subtitles with timings.
Your file stays on your device
Most transcription sites upload your file to their servers. Each minute costs them money, so they cap the free plan.
Here the speech model runs inside your browser tab. The recording is never sent anywhere, which is why there is no limit.
For a cleaner transcript
- Pick the language if you know it. Detection works, but a set language makes fewer mistakes.
- Use Accurate for accents, noisy rooms and names. It downloads more and runs slower.
- One voice close to the microphone gives the cleanest text. Crosstalk and music cause mistakes.
Recording meetings to get the notes?
Cedric joins your Google Meet, Zoom and Teams calls like a teammate. When the call ends he posts the minutes in Slack, and every to-do he heard waits for your approval.
More free tools
See all tools →FAQ
Yes. There is no account, no daily cap and no watermark. Your own computer does the work, so a file costs us nothing.
No. The file is read and transcribed inside this browser tab. The page downloads the speech model once, and your recording is never sent anywhere. Once the model has loaded, it works with Wi-Fi off.
Yes. Drop an MP4, MOV or WebM and the tool transcribes its sound. Download the SRT or VTT file to add subtitles to the video.
Close to 100, including English, French, German, Spanish, Italian, Portuguese, Dutch, Japanese and Chinese. Leave it on Detect, or pick the language for better results.
On a test laptop with no graphics card, Fast took about 10 seconds per minute of speech. A graphics card makes it quicker, a phone makes it slower. The text appears as it goes, so you can read the start before it finishes.
No. You get one stream of text with times. When Cedric takes the notes in a call, he writes down who decided what and who owns each to-do.
Clear speech close to the microphone comes out nearly word for word. Crosstalk, music and strong accents cause mistakes. Accurate catches more of them, with a bigger first download.
Your AI employee is waiting for you
Get started for free100K credits ($50) on sign up. No card needed.