Translate only one application
With a game and Discord both open, choose the game to subtitle its dialogue. Choose Discord to read what your teammates are saying.
Audio source · One app or system audioTurn unfamiliar voices into
subtitles you can follow.
Translate just the game or app you choose. Use cinema subtitles for story dialogue or the classic overlay for team chat.
Five languages · 20 online translation directions · Japanese, Korean and Russian in Beta · English ↔ Chinese offline
VoxGo captures Windows system audio, recognizes speech on your PC, translates it, and sends the result to your game overlay. Scroll through the signal path.
With a game and Discord both open, choose the game to subtitle its dialogue. Choose Discord to read what your teammates are saying.
Audio source · One app or system audiofaster-whisper recognizes speech locally on your PC. VoxGo Lite downloads its recognition model on first use.
He is one shot behind the box.LOCAL ASR: ACTIVEIn v0.6.1, choose one of 20 direct online directions or English ↔ Chinese local translation. The overlay shows the active mode.
Rotate B, don't peek mid. → 转 B,别探中路。MODE: API / LOCALOriginal speech and translation appear in a transparent always-on-top subtitle window. Copy a completed subtitle or mirror it to a phone on your local network.
LIVE SUBTITLE / 转 B,别探中路。OUTPUT: LIVE SUBTITLEChinese, English, Japanese (Beta), Korean (Beta) and Russian (Beta): each source can translate directly to the other four languages.
| Language | Local ASR | Online | Offline |
|---|---|---|---|
| Chinese (Mandarin) | Supported | Other four languages | English ↔ Chinese |
| English | Supported | Other four languages | English ↔ Chinese |
| Japanese (Beta) | Supported (Beta) | Other four languages | Not supported |
| Korean (Beta) | Supported (Beta) | Other four languages | Not supported |
| Russian (Beta) | Supported (Beta) | Other four languages | Not supported |
Whisper recognizes speech on your PC. Your configured OpenAI-compatible provider/model or Google Cloud Translation Basic v2 translates recognized text online using your own API key/service access. Local OPUS supports English ↔ Chinese only; the other 18 directions require online translation. Provider language support and quality vary. No silent fallback or intermediate-language translation.
Beta languages may miss words, mistranslate or respond slowly with short callouts, noise, music or overlapping voices. We will continue improving recognition, performance and subtitle stability.
Speech recognition runs locally in either mode. Translation can use a configured online provider or an on-device model.
v0.6.1 routes 20 online directions through your OpenAI-compatible service or Google Cloud Translation Basic v2. Bring your own API key; language support depends on the provider/model.
Local OPUS is English ↔ Chinese only. Prepare the model to translate offline without an API key. The other 18 directions require an explicitly configured online provider.
Trace every improvement to its GitHub Release.
Recommended for most players. First run guides you through translation and audio tests, then downloads the speech recognition model.
A smaller download for your first run. Includes local Whisper recognition, API or local translation, and the game subtitle overlay.
Includes multilingual Whisper small/base models to reduce the first-run download.
Includes multilingual models and CUDA DLLs for NVIDIA users.
Need every package and checksum? View the Latest Release
Check translation and audio before joining a match. Most subtitle issues come from the wrong Windows output device.
Download Lite, Full, or Full-CUDA and unzip the archive.
Launch VoxGo.exe and complete the first-run setup.
Choose OpenAI-compatible or Google Translation and run the API test.
System audio is the default. On Windows 11, you can choose “Translate only one application” in Settings → Speech. Play some dialogue to check the level meter.
Move and lock the overlay, then use hotkeys or the tray icon in-game.
Share your experience, get help and release news, and discuss what VoxGo should build next.
Announcements, support, and feature ideas.
Join Discord ↗Announcements and discussion in Chinese.
Join KOOK ↗Include version, device, and steps to reproduce.
Open an Issue ↗Browse the code, contribute, and find releases.
View source ↗For discussion, usage questions, and feature ideas, join either community. For reproducible bugs or code issues, open a GitHub Issue.
Short answers for the things people usually ask before trusting an overlay tool.
Recognition can miss words or sentence endings; translation may change entities, numbers or negation. Noise, music and overlapping speech add risk. Latency depends on your CPU, audio segmentation, network and provider. Beta means ongoing improvement; no accuracy percentage, fixed latency or game FPS guarantee is claimed.
VoxGo captures game or Discord system audio. v0.6.1 supports five-language recognition and 20 online directions, with Japanese, Korean and Russian marked Beta.
Yes, for English ↔ Chinese only. Prepare the local OPUS models; no API key is required. Japanese, Korean and Russian directions require online translation in v0.6.1.
Not by default. VoxGo is designed to capture Windows system audio output, such as game voice, Discord, or video audio.
No. VoxGo is a standalone desktop app. It captures audio through Windows devices and displays subtitles in a transparent always-on-top window.
The selected audio device may not match your current Windows output. Use the audio test meter in setup or settings, then choose the loopback device that moves when game or Discord sound plays.
v0.6.1 supports Chinese (Mandarin), English, Japanese (Beta), Korean (Beta) and Russian (Beta), with 20 online directions. See the language support table above.
Lite has no model and no CUDA DLLs. Full includes multilingual Whisper small/base but no CUDA DLLs. Full-CUDA includes those multilingual models and CUDA DLLs for NVIDIA users. English-only models download on demand only after enabling Pure English.
VoxGo keeps the risky parts explicit: no game process injection, system audio capture instead of microphone by default, local Whisper ASR, and user-controlled translation providers.
The overlay is an independent window. It does not patch, hook, or modify game clients.
Whisper transcription runs on your PC. Translation requests go to the provider you configure.
The code is available on GitHub, including packaging scripts, update manifest, and runtime behavior.
For reproducible bugs, the built-in feedback template includes version, device, provider, latency, and log paths. Join the community for usage questions and feature ideas.