Speak, and the box fills with text
Press Start Dictation, allow the mic, and start talking. Words show up as you speak. Rough guesses appear first in a lighter shade, then the recognizer locks them in once it’s confident. The difference from a lot of dictation tools? The transcript here is a real editable textarea. Caught a wrong word? Just click in and fix it like any other text field.
It leans on the Web Speech API baked into Chrome, Edge, and the other Chromium browsers. No upload, no install, no account.
What it does
- Live transcription that grows as you keep talking
- 17 languages and regional accents in the dropdown
- An editable output box, not a locked panel
- Interim results so you see progress before each phrase finalizes
- Copy to clipboard and download as a
.txtin one click - A running word count under the box
Getting set up
Open the page. Pick your language. Hit Start Dictation and approve the microphone prompt the first time. Then talk like you would to a person.
A couple of things help a lot. Use a headset or a decent mic instead of a laptop’s built-in one. Background hum and clattering keyboards drag accuracy down fast. Pause briefly between sentences so the recognizer can chop your speech into clean chunks. And say your punctuation out loud where the engine supports it, since “period” and “new line” often register as the actual marks.
When you stop, the text stays put. Edit it, copy it, or download it. Nothing clears until you tell it to.
Where this comes in handy
Drafting at the speed of talking. Most people speak roughly three times faster than they type. Got a messy first draft swimming around your head? Say it out loud, get the bones down in a fraction of the time, then tidy up the wording afterward.
Hands-free input. Cooking, holding a baby, nursing a wrist that’s tired of typing? Dictation keeps you producing text without touching the keyboard. That matters for anyone dealing with repetitive strain or limited mobility.
Quick meeting and call notes. Keep the tab open and capture the gist while a conversation happens. It won’t beat a pro transcription service for a full record, but for the three decisions that actually mattered, it’s plenty.
Pronunciation feedback. Switch to a language you’re learning and read a passage aloud. If the recognizer mangles what you said, that’s a blunt but honest signal about your accent.
Journaling and brain dumps. Sometimes you just want to vent into a box without staring at a blinking cursor. Talk it out, download the file, move on.
FAQ
Which browsers actually work?
Chromium ones. Chrome and Edge are the most reliable, and Brave and Opera work too since they share the engine. Firefox doesn’t expose speech recognition at all, and Safari’s support is spotty. If you see the “not available” message, switch browsers.
Does my audio get sent somewhere?
In Chrome it can. Chrome’s recognition is partly cloud-based, so it ships audio to Google to transcribe. The page itself stores and transmits nothing beyond what the browser’s API needs. For sensitive material, pick a browser with on-device recognition or skip cloud dictation entirely.
Can I change the language while recording?
Nope. Set the language before you press Start. To switch, stop the session, choose the new language, and start again.
Why does it stop on its own sometimes?
Browsers often time out after a stretch of silence. That’s normal. Hit Start Dictation again and keep going where you left off.
How accurate should I expect it to be?
In a quiet room with clear speech and a good mic, mainstream languages land around 90 to 95 percent. Noise, strong accents, and cheap microphones pull that number down quickly.
Can I fix mistakes without retyping everything?
Yep. The transcript is a normal editable field. Click anywhere, correct the slip, and carry on. Just do your edits while dictation is stopped so new speech doesn’t overwrite them.