Audio Translator

Real-time translation of any audio in your browser: live subtitles in 39 languages. Runs on your computer, no account needed.

Live translate audio from any Chrome tab. Audio Translator is a free Chrome extension that turns any sound that plays in the browser into real-time translated subtitles (live captions). Talk on Zoom, Google Meet or Microsoft Teams with people who speak another language and read what they say in your language. Translate YouTube videos, movies and series, even ones with no captions. Or follow a Twitch stream or a podcast, or sit through an online lecture or webinar in a language you don't know. It understands speech in 33 languages and writes subtitles in 39, usually less than a second behind the speaker. No account and no minute limits. Nothing is sent to a server: local models do all the work right on your computer, and once the language pack is downloaded it works offline. Only the video or call itself needs the internet. Pick the same language on both sides and you get live transcription (speech to text) instead. Translate Zoom, Google Meet and Teams calls live Open the meeting in the browser, not the desktop app (Zoom, Meet and Teams all have a web version), and the language barrier with colleagues or clients mostly goes away. Any pair of the supported languages works, and nobody else in the call has to install anything. How to use it 1. Open the tab with the call, video, movie or stream and click the Audio Translator icon. The side panel opens. 2. Choose the spoken language (Auto usually guesses right) and the language you want to read. 3. The translation shows up while the sentence is still going, and finished lines scroll up. To see them on the video instead, press Overlay at the bottom of the panel. What you get • Subtitles sit in the side panel, or in an overlay on the video (full screen too). You can drag and resize it and pick the text size. The background goes 90, 60, 30% or off, and with it off you see only the text on top of the picture. • You can also show what was actually said: above each translated line you see the same line in the speaker's own language. • Autosave writes a transcript of every session to a text file in Downloads/audio-translator, which is basically your lecture or meeting notes. • The panel interface comes in 18 languages. Languages Speech recognition: English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Czech, Slovak, Slovenian, Croatian, Bulgarian, Romanian, Hungarian, Greek, Lithuanian, Swedish, Danish, Finnish, Norwegian (Bokmål and Nynorsk), Russian, Ukrainian, Turkish, Arabic, Hebrew, Hindi, Chinese (Mandarin), Japanese, Korean, Thai, Vietnamese. Subtitles: all of the above (Norwegian counts as one here), plus Bengali, Indonesian, Kannada, Marathi, Tamil, Telugu and Traditional Chinese. Limits Clear speech works best; loud music or three people talking over each other will cost accuracy. It's a big download, about 450 MB, because the model is inside. The first start takes a few seconds while the model loads. Each new language pair downloads its language pack the first time you use it. You need desktop Chrome 138 or newer. No graphics card is required: speech recognition runs on the processor. If you want it to run on your graphics card instead, set Compute to GPU in settings. FAQ How is it different from YouTube's auto-translate? YouTube translates existing captions, and only on YouTube. Audio Translator works from the sound itself, so it doesn't care whether captions exist or which site is playing. Why does it ask for extra permissions? Site access is only needed for the overlay, to draw subtitles on the page. Download access is only for Autosave. The side panel needs neither. Page content is not read or sent anywhere.