GAMING & STREAMS

Play games that were never localized for you.

VoxisLive translates game audio, cutscenes and live streams into a spoken voice in real time — no virtual audio cable, no file patches, no mods. Keep your eyes on the screen and your hands on the controller.

The untranslated catalog problem

The global catalog of Japanese, Korean and Chinese games dwarfs what Western publishers localize. Thousands of RPGs, visual novels and action titles remain untranslated — indefinitely, or for years after release. Fan patches cover a fraction of the catalog, lag behind updates, and require technical setup.

VoxisLive approaches the problem from the audio output rather than the game files. If the game speaks its dialogue aloud, VoxisLive translates that dialogue and speaks it back to you, in real time, while you play. No patch, no file modification, no ROM editing.

How it works with games

WASAPI loopback reads the audio your sound card is playing — the same low-level API Windows uses internally. The moment a character speaks or a cutscene begins, on-device speech detection isolates the line and the translation comes back as a spoken voice, fast enough that voiced dialogue usually translates before the character animation ends. Unvoiced text is not translated — VoxisLive is an audio tool, not an OCR engine.

Why spoken beats subtitles in games

Reading subtitles pulls your gaze to the bottom of the screen — costing positional awareness, reaction time and immersion in any game with moment-to-moment demands. Audio is processed through a different cognitive channel than vision: you hear the translated line without looking away, and you can even keep the original voice performance underneath.

No anti-cheat surface, no drivers

Virtual audio cables insert a fake audio device that some games, launchers and anti-cheat systems flag as unusual software. VoxisLive installs no driver, injects no code into game processes and creates no virtual device — it reads from the same WASAPI API Windows itself uses, entirely outside the game process.

Streams too

VoxisLive doesn't distinguish between a game, a Twitch stream or a YouTube video — all produce audio through your sound card. A Japanese Twitch stream, a Korean esports broadcast, a Spanish gaming channel: select your system output, set the language, press play. Streaming yourself? VoxisLive's translated output goes to a separate channel, so OBS captures your stream audio unchanged.

What works

Any Windows game with voiced audio, regardless of engine, launcher, DRM or store — Steam, Epic, GOG, Microsoft Store or plain executables. That includes VRChat and voice-chat-heavy multiplayer, where teammates' voice chat is captured the same way. Languages: 79, in either direction — Japanese→English, Korean→English, or any supported pair.

FAQ

Common questions

01Can VoxisLive translate a game with Japanese voice acting but no English dub?
Yes, as long as the dialogue is voiced. VoxisLive captures whatever audio Windows is playing, detects the speech, and speaks the translation in your language. No game modification, patch or file access is needed.
02Will it conflict with anti-cheat software?
No. VoxisLive does not inject code into game processes, installs no driver, and creates no virtual audio device. It reads from the Windows WASAPI loopback API entirely outside the game process, presenting no surface that anti-cheat systems monitor.
03Can I use it while streaming on Twitch with OBS?
Yes. VoxisLive doesn't alter your audio routing; OBS captures the same devices it always has. The translated voice goes to a separate output (e.g. your headphones), so it doesn't appear in your stream unless you route it there.
04How many minutes does gaming use?
Minutes are only consumed on detected speech, so it depends on how dialogue-heavy the game is. A voiced JRPG uses more than a game with sparse cutscenes. Prepaid packs from 80 to 2400 minutes cover every pattern — and the BYOK build is free with your own key.
Free to try · 10 minutes on us

Hear every language, in real time.

Runs on Windows 10 and 11 — no drivers, no setup ritual, no bot in your call.