Packages & APIs: Native W3C Web Speech API (SpeechRecognition / webkitSpeechRecognition), HTML5 Web Audio API, and client-side SRT subtitle formatters.

⚖️ Legal notice: this tool is provided "as is", without warranty of any kind. That includes accuracy and fitness for any particular purpose. It uses the browser's Web Speech API — nothing else. In Firefox, recognition runs on your device. In Chrome and Chromium browsers, the built-in engine sends audio to Google's speech service to transcribe it — that is how the browser's engine is built. We add no storage, no logging and no upload of our own. No account is created, and nothing is retained by us.

How this tool works

Press start, speak, press stop. The browser's Web Speech engine turns what it hears into text while you watch, and the box above holds the result — as timed .SRT captions or as plain text you can copy or save. Nothing is stored by us at any point; when you close the tab, what you did not save is gone.

Who it is for

Talking is easier than typing for many people. Hands tire. Motor control makes keyboards slow. A thought moves faster than fingers can chase it — and for speech that is variable, speaking a note while it is fresh can beat fighting a keyboard after the moment has passed. This page exists for exactly those moments: notes, drafts, ideas caught in motion. It is built as a single-speaker tool for short bursts, not a meeting recorder, and it does not pretend otherwise.

How to use it well

Work in short bursts. A few sentences at a time keeps the engine accurate and the editing box human-scale.
Speak your punctuation. "Period", "comma" and "new line" mostly land as written marks — worth knowing before the first surprise.
Read before you save. The engine is your browser's, not ours; it mishears names and jargon like any speech tool. The text is a draft until you have looked at it.
Quiet room, steady distance. The recognition quality tracks the microphone's life story more than anything this page controls.

What this is not

Not speech therapy, and not a language tool — it transcribes; it does not treat, teach or evaluate anything.
Not an accuracy promise. Results track the engine your browser ships, which varies by browser, language and microphone. "As is" is the honest word for it.
Not private in Chrome. The most important honest line on this page lives in "Where your words go" above — read it before dictating anything sensitive.
Not a transcription service. No storage, no history, no accounts — and no recovery if the tab closes before you save.

Questions people ask

Where do my recordings go?

In Firefox, the recognition runs on your device. In Chrome and other Chromium browsers, the built-in engine sends audio to Google's speech service to be transcribed — that is how the browser's engine is built. We add no storage, logging or upload of our own, and nothing is retained by us either way. If you need the audio to stay on your device, Firefox is the honest choice.

Why .SRT?

It is the standard subtitle format, and it happens to be a good plain-text carrier for speech with timing: a line, a timecode, the words. If you do not need timing, save as .TXT instead.

Does it work offline?

The page runs in your browser, but the speech engine belongs to the browser, not to this page — and in Chrome that engine needs the network to reach Google's service. Treat it as an online tool with an offline-shaped output.

Where to go next

My Board — for when the hard direction is the reverse: getting words out when speech is unreliable.
Ask & Wonder — a phrase board with gentle facts, for curious minds who would rather listen than read.
The privacy page — the exact rules for every tool on this site, including this one.