How to get live captions for VRChat and Discord voice chat
Voice chat has no subtitle button. If you are deaf or hard of hearing, if the room is loud, or if the people around you speak a language you do not know, you need something that turns the voices you hear into text on your screen.
There is a free place to start on Windows 11, and there are tools that add translation. All of them share limits worth knowing before you spend an evening on setup.
Start with Windows 11 live captions
Windows 11 (version 22H2 and later) includes a feature called live captions. Microsoft says it turns “any audio that passes through your PC” into captions, and that covers a Discord call or VRChat running on your desktop.
To turn it on:
- Press the Windows logo key + Ctrl + L. You can also go to Settings > Accessibility > Captions and turn on the Live captions toggle, or use Live captions in quick settings.
- The first time, agree to voice processing on your device and let Windows download the language files.
- Select the Settings button in the captions window to place it above the screen, below the screen or overlaid on the screen.
- To caption a different spoken language, select the Settings button, then Change language.
Microsoft says all processing happens on your device, that audio and captions never leave it, and that captions are not stored.
The spoken languages Microsoft lists, as of September 2026, are Chinese (Simplified and Traditional), Danish, English, French, German, Italian, Japanese, Korean, Portuguese and Spanish.
Does it translate?
Only on some hardware. Microsoft says translation is available on Copilot+ PCs running Windows 11 version 24H2 or later, from over 40 languages into English and from 27 languages into Chinese (Simplified).
On any other PC you get captions in the language being spoken. That helps if you read the language better than you follow it by ear, but not if you cannot read it.
What it cannot do
- Microsoft says lyrics in music are not reliably detected, and that resource-intensive apps might delay the captions. A game is one.
- Microsoft’s description mentions no speaker labels, so do not expect the captions to say who is talking.
- It is a window on your Windows desktop. It is not part of what VRChat draws inside a VR headset.
Live captions for Discord calls
We found no caption feature for voice in Discord’s documentation as of September 2026. On a Windows 11 PC, live captions are the simplest free answer: start the call, press the shortcut, and set the caption language to the language being spoken.
Some voice bots also post text: NeuroVox says it sends its translation as a text message as well as speaking it. Bots only work in servers, and someone with permission has to add them. Our guide to translating Discord voice calls compares that route with the others.
On an Android phone, Live Caption can caption speech on the device, including video calls. Google says it “isn’t intended for calls with more than one other person” and may be unavailable in some calling apps.
VRChat subtitles
We found no built-in voice captions in VRChat as of September 2026. A “Live Translation” request posted on VRChat’s feedback board in April 2025 shows no status. Subtitles come from two directions.
From the speaker. A player can run a tool that writes their own speech, or a translation of it, into the chatbox above their avatar. You cannot switch it on for someone else, but you can ask friends to use it. See chatbox subtitles over OSC.
From your side. You capture the game audio you hear and caption it. Windows live captions do this, without translation unless you have a Copilot+ PC. The tools below add translation.
Translated subtitles: tools that caption system audio
These tools listen to what your PC plays and show it as text in your language. Details are from each tool’s own pages, as of September 2026.
| Tool | Where the work happens | Where you read it | Good to know |
|---|---|---|---|
| Windows 11 live captions | On your PC | Desktop caption window | Built in. Translation on Copilot+ PCs only |
| VRCT | On your PC with its default translation engine, which works offline. Cloud engines are optional | Its chat history, and an overlay mode for SteamVR | Free. It transcribes all audio from the device you select |
| Whispering Tiger | On your PC (“runs 100% locally”) | A web browser view, or over OSC | Free and open source. Its README says the models can take up to 20 GB |
| VRCLT | A cloud engine you choose, with your own API key | Its dashboard and a SteamVR subtitle overlay | Free app, with a Discord mode |
| TalkFerry Understand | Cloud | A desktop panel that stays on top of the game | Paid minutes after 5 free minutes. No API key or GPU |
If you play in a headset, VRCT and VRCLT have an advantage: both can show text inside SteamVR. TalkFerry’s panel is a desktop window and is not visible in a headset by itself.
TalkFerry’s Understand mode listens to the game or voice-chat audio and shows live subtitles in your language. It can also speak the translation in your headphones. It processes everything it hears, so it uses minutes faster than translating your own voice. The cleanest setup sends the game’s audio to a second virtual cable that you pick as the “room line” in Settings. Without one, it listens to your speakers and pauses while it talks.
Because it runs in the cloud, the audio goes through TalkFerry’s server to Google Gemini. TalkFerry does not record or store audio or transcripts. While it runs on Gemini’s unpaid tier, Google may use the content to improve its products and human reviewers may read or listen to it, so do not use it for anything confidential. The tools that work on your own PC are the better choice when privacy comes first.
Limits that apply to every caption tool
Several people talking at once. When two voices overlap, words get dropped or merged. Microsoft notes that if you caption your own microphone too and speak over the other person, you only see captions for the other person. A busy public instance or a full voice channel is hard for any of these tools.
No names. A tool that listens to your PC’s audio output gets one mixed stream, so it cannot know which account a voice belongs to. TalkFerry cannot tell speakers apart at all. VRCLT’s Soniox engine numbers the voices it detects, and its README is clear that these are labels for the session, not VRChat identities.
Music and noise. World music, video players and other apps go into the same stream. VRCT’s docs warn that its speaker transcription picks up all audio from the selected device and advise keeping other desktop audio down.
Delay and odd words. Captions trail the voice, and usernames, world names and in-jokes are the words most likely to come out wrong. Our explainer on how voice translators work covers why.
A note for deaf and hard-of-hearing players
Captions can make voice chat easier to follow, and Microsoft names people who are deaf or hard of hearing among those live captions are meant to help. They are still automatic captions: they drop words, lose overlapping speech, do not name speakers and do not describe sounds that are not speech.
A few things help in practice:
- Try Windows live captions first. They cost nothing and keep audio on your PC.
- Prefer small groups and quiet worlds, where one voice at a time is realistic.
- Ask friends to type, or to use chatbox text in VRChat, for anything that matters.
- Do not rely on automatic captions for anything safety-critical.
Where to start
If you read the language being spoken, turn on Windows live captions and see whether that is enough. If you need translation and want everything to stay on your PC, try VRCT or Whispering Tiger. If you would rather not run models or manage an API key, TalkFerry’s Understand mode is one option, with the cost and privacy trade-offs above.
Sources
- Microsoft Support: Use live captions to better understand audio
- Google: Live Caption on Android
- VRChat feedback board: Live Translation request
- VRCT docs: Speaker to Text, Translation Engines and Overlay Mode (VR)
- Whispering Tiger on GitHub
- VRCLT on GitHub
- NeuroVox
- Discord developer documentation: bot users
Frequently asked questions
Does Discord have live captions for calls?
We found no built-in captions for voice channels or calls in Discord's documentation as of September 2026. On a Windows 11 PC, the system's live captions can caption a Discord call, because they work on any audio the PC plays.
Does VRChat have subtitles?
We found no built-in voice captions in VRChat as of September 2026. Players can show their own speech as chatbox text with a third-party tool, and listeners can caption the game audio with Windows live captions or a subtitle tool.
Can Windows live captions translate?
Only on Copilot+ PCs running Windows 11 version 24H2 or later, where Microsoft says they translate from over 40 languages into English and from 27 languages into Simplified Chinese. On other PCs you get captions in the language being spoken.
Can captions show who is speaking?
Not by name when the tool listens to your PC's audio output, because every voice arrives in one mixed stream. A few tools number the voices they detect, but those numbers are guesses for the current session and are not tied to accounts.
Are live captions good enough for deaf and hard-of-hearing players?
They help many people follow a conversation, but they miss words, lose overlapping speech and do not name speakers. Treat them as an aid next to text chat, not as a replacement for it.
We make TalkFerry, one of the tools mentioned here. Software changes quickly: check each tool’s own page before you rely on a detail. Found a mistake? Write to support@talkferry.com.