Both directions: they hear your voice in their language
Can the other person hear me in their language? Yes. Meeting mode runs two translations at once. Incoming: the meeting audio is captured from your Windows sound output and spoken to you in your language through your headphones. Outgoing: your microphone is translated into the meeting language and spoken into a virtual microphone (a free driver such as VB-CABLE), which you select as your mic in Zoom, Teams or Meet. The other participants hear a natural voice saying your words in their language, a few seconds behind you — no interpreter line, no bot, no second participant.
What it is not: it does not clone your own voice on the outgoing side, and it does not remove the delay of simultaneous interpretation — expect a short lag, the same span a human interpreter needs. Incoming translation is driverless; only the outgoing leg needs the virtual microphone. Meeting mode works with 79 target languages and every conferencing app that runs on Windows. Ideal for international sales calls, cross-border interviews and support sessions.
Both pickers are target languages — you never set the spoken one
You only ever choose two things: the language you want to hear, and the language the other side should hear. Whatever language is spoken on the other end, you hear it in the one you chose — no source language to declare, no extra setting to find. VoxisLive handles the rest.
In Meeting mode neither selector asks what language you speak. One sets what you hear (“I hear in”), the other sets what the other side hears (“They hear in”). The spoken language is detected automatically, sentence by sentence, in both directions.
So whatever language you speak, it reaches them in the language you picked — and even if the other side is multilingual, or three languages are spoken around the table, you keep hearing the single language you chose. Switch mid-call from English to Spanish and the person on the other end still hears their own language: nothing to change, nothing to re-pick.
The problem with meeting translation today
Sitting in a call conducted in a language you don't fully understand is exhausting: you catch fragments, miss the nuance of a question, and spend half the meeting reading machine subtitles that scroll past before you can parse them.
The standard industry answer — a translation bot that joins the call as a participant — creates its own problems. Your colleagues can see it. Your host may not allow third-party attendees. And you've handed a recording of your conversation to a third-party server the moment the bot connects.
VoxisLive takes a different approach: it sits on your Windows machine, listens to what your speakers are already playing, and speaks the translated audio back to you — privately, without ever touching the call itself.
How does it translate Zoom, Teams and Meet audio?
VoxisLive reads audio directly from your sound card's output stream via WASAPI loopback. When someone speaks, the audio travels from the conferencing app to your speakers as it normally would; VoxisLive intercepts that stream at the OS level, isolates speech with on-device detection, and returns a spoken translation within seconds.
Because capture happens at the OS audio layer — not inside the conferencing app — the process is invisible to every participant. No bot joins, no extra attendee appears, no notification fires. Zoom, Microsoft Teams, Google Meet, Webex, Slack Huddles and Discord all work identically.
Why bot-free matters
Meeting bots used by competing translation services connect a second participant to your call: it has a microphone feed, generates a recording, and is visible in the meeting. Several enterprise platforms flag or block unrecognized bots by default, and many corporate security policies prohibit third-party bots outright.
With no bot, none of that arises: no IT whitelisting, no new data processor for legal review, no unknown attendee for the host to admit. For regulated industries, healthcare, legal proceedings or any genuinely confidential meeting, bot-free translation is not a convenience — it is the minimum viable standard.
Getting started takes under five minutes
- Install VoxisLive from the Microsoft Store. On Linux, install it from the Snap Store instead:
sudo snap install voxis. - Pick the Meeting scenario and choose your microphone.
- Set I hear in and They hear in — both are target languages, and the spoken one is detected automatically (79 languages).
- Start the session, then join your meeting as you normally would.
The app runs quietly alongside your call. Want to see exactly how the data flows? The engine's capture and session code is published on the source-available excerpt on GitHub. See how it works and pricing for details.