About this tool
Show adjustable live captions using your browser's own speech engine.
Live Caption Overlay turns speech picked up by your microphone into large on-screen captions using the browser's own Web Speech Recognition API, in continuous mode with interim results so words appear while the sentence is still being spoken. Caption text sizes from 18px to 54px on a high-contrast panel, and you can set the recognition language to English (India, US or UK) or Hindi before starting. It is a quick accessibility aid for a room or a call — not a transcription service, and it works only in browsers that expose the speech-recognition API, which today mainly means Chromium-based ones.
Open Live Caption Overlay on AltFTool — it loads instantly in your browser.
Choose a Recognition language — English (India), Hindi (India), English (US) or English (UK) — and set the "Caption size" slider anywhere from 18px to 54px in 2px steps; the language selector locks once recognition is running.
Press "Start captions" and grant the browser's microphone permission; recognition runs continuously with interim results, so words appear before the sentence is finished.
Captions fill the high-contrast panel, which is an aria-live polite region so screen readers announce new text; "Stop captions" ends the session and Clear wipes the transcript, which is never saved or exported.
Interim results are shown as they are recognised and then replaced by the final text, so the caption keeps pace with the speaker.
Caption text scales from 18px up to 54px in 2px steps on an inverted high-contrast panel, readable from a distance.
The caption area is an aria-live polite region, so assistive technology picks up new text without you refocusing it.
Only browsers that implement the Web Speech Recognition API — SpeechRecognition or webkitSpeechRecognition — which in practice means Chrome, Edge and other Chromium browsers. If the API is missing, the tool says so instead of failing silently, and no captions will appear.
That depends entirely on your browser's speech engine, not on this page. Several browsers send audio to a cloud recognition service, so do not use live captions for confidential conversations unless you know your browser recognises speech on-device.
Four options are offered: English (India), Hindi (India), English (US) and English (UK). The language must be chosen before you press Start, because the setting is locked while recognition is running — stop, switch, and start again to change it.
No. Text accumulates on screen while the session runs and disappears when you press Clear or leave the page — there is no file, no history and no export. Copy anything you want to keep before closing the tab.