Speech to Text — Voice Transcription

Press Start, speak, and your words appear in the transcript box in real time. Speech to Text runs the browser's built-in SpeechRecognition API, so nothing is uploaded and no account is needed — the audio never leaves the tab. When you stop, the transcript is a normal editable text field: fix a misheard word, move a sentence, then copy or download the result.

Twenty languages are available in the selector. English mode in Chrome automatically converts spoken punctuation commands: saying “comma” inserts , and “period” inserts . without any chip. For other languages, or when you want a dash or paragraph break, the eight punctuation chips at the bottom of the tool insert the mark at the end of the current transcript and resume from there.

Getting Your First Transcript

1. Choose a Language from the selector. The browser will use the matching recognition engine; accuracy depends on the voice models installed in your OS.
2. Press Start and grant microphone permission when prompted. The button turns terracotta-orange and pulses to confirm it is live.
3. Speak at a normal pace. Confirmed words appear in the transcript in black; the current in-progress phrase appears in grey below while the engine is deciding.
4. Click anywhere in the transcript box to edit a misheard word mid-session, without stopping the microphone.
5. Press Stop, then Copy to copy to the clipboard or Download .txt to save the file named transcript.txt.

Where Your Data Goes

The audio stream goes to the speech recognition engine your browser designates — in Chrome and Edge that is Google's cloud service, in Safari it is Apple's, and in each case it is governed by the browser's own privacy policy rather than this site's. The transcript text itself is processed and stored only in your browser tab. Closing or reloading the page clears it entirely; there is no server-side storage and no account to sign into.

Edge Cases: Browser Support and Silence Timeouts

• Chrome, Edge, and Chromium-based mobile browsers have full support. Safari 14.1+ works on Mac and iOS. Firefox does not implement SpeechRecognition and will show an unsupported notice.
• Some browsers stop recognition automatically after roughly 6–10 seconds of silence. The tool restarts the session automatically to keep the microphone live, but a very long pause may result in the final word of a sentence being held as interim — press Stop then Start to flush it.
• Changing the Language selector while the microphone is active is disabled. Stop the session first, change the language, then start again. The existing transcript is preserved.
• On mobile, the browser may request permission again each session. This is a device-level permission setting, not something the page controls.