Speech to text: dictate for free, right in your browser
Click the microphone, talk, and your words turn into text you can copy anywhere. The speech model runs on your device, so your voice never leaves it.
- No upload
- No sign-up
- 90+ languages
Click the mic and start talking
Your browser will ask for microphone access once.
- 0 words
Everything runs locally: your microphone audio is processed in your browser and never uploaded. Only the speech model (about 77 MB) is downloaded once from Hugging Face.
Want this in every app, not just a text box?
Ownvox dictates straight at your cursor on Mac and Windows: one hotkey, any app, offline by default.
How it works
- 01
Allow the microphone
Click the mic once. Your browser asks for permission; the audio stays in the browser's memory and is never sent anywhere.
- 02
Speak in sentences
Talk as you normally would. Each time you pause, the sentence you just said is handed to the Whisper model on your device.
- 03
Text appears, copy it
A moment later the sentence shows up in the text box. Fix a word if you like, then copy or download the result.
Why dictate locally?
Most online dictation tools stream your microphone to a server. This one cannot: the model runs on your machine, so there is nothing to upload, store or train on.
Your voice stays here
Notes, emails, drafts of things you have not said to anyone yet. All of it is processed on your device and gone when you close the tab.
No sign-up, no minutes
No account, no free-minute counter, no watermark. Dictate for as long as you want.
Works offline
Once the model is cached, dictation needs no connection. Switch off Wi-Fi and check for yourself.
Browsers, languages and accuracy
Browsers and devices
Works in current Chrome, Edge, Firefox and Safari on desktop and on most Android phones; on iPhone, Safari works but is slower. A computer with a supported GPU (WebGPU in Chrome or Edge) turns sentences into text almost instantly; on the CPU path each sentence takes a few seconds.
Languages
Whisper understands more than 90 languages. Pick the language you will speak before you start; it tells the model what to expect and matters most for short sentences and accents. The tool transcribes, it does not translate.
Accuracy
This page uses Whisper "base". In a quiet room with clear speech it is very readable; strong accents, background noise and specialist terms bring the quality down. Punctuation is added automatically, and you can edit the text at any time.
Frequently asked questions
You're not slow.Your keyboard is.
One hotkey, one voice, one cursor. Text lands where you're working, instantly.
7 days free, then from €9.99/mo · cancel anytime, effective at period end