Live captions for any stream or VTuber broadcast playing outside your browser—TikTok Live, Bilibili, Kick, or any desktop app.
A Japanese VTuber streams on Bilibili through the desktop client. TikTok Live runs in the TikTok app on your PC. Someone hosts a Kick stream you watch in a third-party player that embeds chat and video. These applications run outside Chrome, so browser-based translation extensions designed for YouTube or Twitch tabs cannot access the audio. Translext uses your computer's system audio sharing to listen to any desktop application and display live translated captions in a floating window over the stream.
Most VTuber clip channels and highlight accounts rely on manual subtitling after the fact, which means you miss live moments entirely. Browser extensions work only when the stream plays in a Chrome tab; native apps, standalone players, and alternative platforms remain inaccessible. Translext bridges that gap by capturing sound from any running program on Windows, ChromeOS, or macOS 14.2 and later. The captions appear in real time, though rapid-fire commentary, overlapping game sound effects, and background music will reduce recognition accuracy.
Open the Translext extension and select Computer audio as your source. Chrome will ask you to choose what to share; pick the option labeled Entire screen—despite the name, this is the only way Chrome can access system sound. Before clicking Share, turn on the checkbox labeled Share system audio or Share tab audio, depending on your browser version. Once you confirm, Translext immediately disables the screen video feed and listens only to the audio stream. A small caption window appears on your desktop, floating above the stream app. You can drag it to any corner or edge where it won't block the VTuber's face or the live chat panel. The original application audio continues playing normally underneath the captions.
Live captions appear within a second or two of the spoken words, but the delay depends on sentence pacing and how quickly the speaker finishes a phrase. VTubers who talk quickly or interrupt themselves mid-sentence will cause captions to lag slightly as the system waits for enough context to translate accurately. Background music, game sound effects, and on-screen alerts can interfere with speech detection, especially during high-energy gaming segments or musical performances. Translext translates only spoken audio; it does not read or translate the live chat messages scrolling in the app. If you need to follow viewer comments or Super Chat donations, you'll still need to read those in the original language or use a separate chat translation tool alongside Translext.
Computer audio works with any desktop application that produces sound: the TikTok or Bilibili desktop clients, standalone Kick or Rumble players, third-party stream viewers, and even VTuber streams embedded in Discord watch-togethers. If the app runs on your Windows, ChromeOS, or macOS 14.2+ machine and plays audio, Translext can listen to it. On macOS, you must grant Chrome permission under System Settings → Privacy & Security → Screen & System Audio Recording before the browser can share system sound. Linux is not supported. Spoken translation works in Computer audio mode as well: Translext removes its own voice from what it hears, so the translation is never picked up again. For streams playing directly in a Chrome tab—such as YouTube Live or Twitch in a browser—use the Browser tab audio option instead, covered on separate platform pages.
Yes, but you should use Browser tab audio rather than Computer audio. When the stream plays in a Chrome tab, Translext can listen directly to that tab's sound without requiring screen sharing. Browser tab audio also supports spoken translation, where Translext reads the translated captions aloud in a natural voice while turning down the original audio underneath. Computer audio is for streams that play in standalone apps or non-browser software. If the platform offers both a web player and a desktop client, the web player usually gives you more features.
Game audio—gunfire, explosions, music, crowd noise—competes with the VTuber's voice for recognition priority. Translext receives a single mixed audio stream from your computer, so loud sound effects can drown out quieter speech or confuse the speech detection model. Fast talkers who layer sentences on top of game callouts or who shout over boss-fight music will see lower caption accuracy. Pausing the game or lowering the game volume relative to voice in the stream app's audio settings can improve results, though you won't control those settings yourself if you're watching someone else's broadcast.
No. Translext listens only to audio. When you share your screen using Computer audio, the extension immediately turns off the video feed and uses only the sound channel. On-screen text, chat messages, donation alerts, and graphical overlays are not captured or translated. If a VTuber displays Japanese text on screen or reads a viewer comment aloud, you'll get a translation of the spoken reading but not the written text itself. For written content, you would need optical character recognition software or a separate screen translation tool running alongside Translext.
The caption window remains visible and stays on top of other windows even if you minimize the stream app or switch to a browser or document. Translext continues listening to system audio as long as the screen-sharing session is active, so captions keep updating whenever the stream produces speech. If you close the stream app entirely or mute your computer's master volume, the audio stops and captions freeze. You can manually close the caption window at any time, and you can restart Computer audio mode if you accidentally stop the sharing session or need to reposition the window.
Computer audio captures all system sound mixed together, so if the stream and Discord both play simultaneously, Translext will attempt to transcribe and translate everything it hears. This creates two problems: voices and audio from different sources overlap, reducing transcription accuracy, and the captions will jump between languages or contexts without indicating which speaker or app produced each phrase. For cleaner results, listen to one audio source at a time. If you need to translate your Discord conversation separately, use Microphone mode for your own speech or ask others to join a browser-based call where Browser tab audio can isolate that tab's sound.
Translated captions and optional audio translation for online meetings, video calls and videos — in Chrome or any app on your computer. No bots to invite, nothing to upload.