Translating a Zoom call, in both directions, without adding anyone to it

Watch the full setup (10:05) — subtitles in 23 languages, voice-over in 9

This page is the Zoom-specific version of Meeting mode. The mechanism is the same for every conferencing app; what differs between them is which menu you pick the microphone and the speaker from, and Zoom's menus are the ones described here.

What actually happens during the call

Meeting mode runs two independent translation sessions at the same time, one per direction. They do not share a connection and neither one waits for the other.

• Incoming — the audio Zoom is playing through your sound device is captured, translated, and spoken back to you in the language you picked. This is the half that works with no extra setup.
• Outgoing — your microphone is translated and the translated voice is played into a virtual microphone, which you then select inside Zoom as your mic. This is the half that needs one free driver installed first.

Both directions run a few seconds behind the speaker, the way a simultaneous interpreter does. That span is part of interpreting, not a setting you can turn down.

Nothing joins the meeting

VoxisLive does not connect to Zoom. It captures at the operating-system audio layer — the same place your speakers read from — so the application producing the sound is irrelevant to it. A Zoom call, a Zoom webinar you are only listening to, or a Zoom window inside a browser are all just audio as far as the capture is concerned.

Practically, that means three things Zoom users usually ask about:

• No participant is added. There is no bot, no second attendee for the host to admit, and nothing in the participant list.
• No calendar access is requested. VoxisLive never asks to connect your Google or Outlook calendar, because it does not need to know that a meeting exists — you start it yourself, when you want it.
• No Zoom account link, no app install inside Zoom. Nothing is added to your Zoom account or your organisation's app marketplace.

If your workplace blocks third-party meeting bots, that policy is not what you are working around here — there is no bot to approve in the first place.

Incoming: hearing the other side

In Meeting mode the capture is pinned to the driverless path. It reads the system mix while excluding VoxisLive's own process, which is what keeps the translated voice out of the capture — otherwise the app would hear its own output and translate the translation. That exclusion is not a setting; it is how the meeting capture is built, and it is why Meeting mode does not offer the other capture routes some users know from Video mode.

Because the incoming direction reads your sound output, it does not depend on your microphone at all. If the microphone half fails for any reason, you still hear the other side.

Outgoing: being heard in their language

For the other participants to hear you in their own language, the translated voice has to reach Zoom somehow — and Windows only lets an application send audio into a call through a microphone device. So you need a virtual microphone: a free driver (VB-CABLE is the usual one) that presents itself to Windows as a normal input device while actually playing back whatever another app feeds into it.

VoxisLive plays your translated speech into that virtual device; Zoom listens to it as if it were a headset. On Windows, Meeting mode will not start without a virtual microphone installed: the app says so before the session begins, rather than leaving you in a call with one direction missing. If you only need to understand the other side, Video / Game mode does that with no extra setup. On Linux, the app creates its own virtual microphone.

Setting it up in Zoom

Zoom keeps its audio devices in one place. Open Settings (the gear icon at the bottom left of the Zoom Workplace window), then Audio. During a call you can reach the same dropdowns faster with the ^ arrow next to the Audio button in the meeting toolbar; that menu also lists Select a microphone, Select a speaker and Microphone modes.

1. In Zoom, set Speaker to the headphones you actually listen on. Not the virtual cable, and never CABLE Input or CABLE In 16ch — see the warning below. If you switch headphones later, set Zoom's Speaker again: Zoom keeps the device you picked.
2. In Zoom, set Microphone to CABLE Output (VB-Audio Virtual Cable) if you want the outgoing direction — despite its name, that is the cable's microphone side; never CABLE Input. Pick it by hand, and set your real microphone back after the call. Leave it on your real microphone if you only need the incoming direction.
3. If you want the outgoing direction, in Zoom open Settings › Audio › Microphone modes and choose Original sound for musicians; leave High-fidelity music mode, Echo cancellation and Stereo audio unticked. Also turn off Automatically adjust microphone volume. Zoom's microphone is now the translated voice, and Zoom's noise removal can cut off the start of words, so let the voice pass through untouched. While it is on, Zoom shows a music-note icon at the top of the meeting window. After the call, switch back to Noise removal (default) and turn the automatic volume adjustment back on.
4. In VoxisLive, pick the Meeting scenario and select your real microphone as its input — that is the voice it translates.
5. Set the two languages: the one you want to hear, and the one the other side should hear.
6. Start the session, then join the Zoom call as you normally would.

Once you are in the meeting, open the ^ next to Audio and confirm: the microphone is CABLE Output (VB-Audio Virtual Cable) and the speaker is your headphones.

The order matters slightly: install the virtual cable before starting Zoom, or Zoom's device list will not show it until you restart the app. Zoom also remembers devices per profile, so it is worth checking the dropdowns once after any Windows update that reshuffles audio devices.

Headphones, and the one cable mistake that creates a third voice

Use headphones. If the meeting audio and your translated voice play out of open speakers, your real microphone picks them up and sends them back into the call. On Zoom this is the familiar echo problem, made worse by the fact that there are now two voices in the room instead of one.

The more confusing failure is a routing one. If you set Zoom's Speaker to the same virtual cable that VoxisLive is feeding your translated voice into, that cable becomes both the way sound leaves Zoom and the way sound enters it. The call audio and the translated voice circle back into the meeting, and participants hear what sounds like a third person on the line, half a beat behind everyone else.

• The virtual cable belongs in one Zoom dropdown only: Microphone.
• Zoom's Speaker should always be a real output device — your headphones.
• If you hear yourself, or the other side reports an extra voice, check those two dropdowns before anything else.

What this is not

• It is not a Zoom integration. It does not read the participant list, the chat, the meeting title or the recording, and it has no knowledge of the call beyond the sound it produces.
• It does not remove interpreting delay. Both directions run a few seconds behind live speech.
• It does not make the outgoing voice sound like you — the other side hears a synthetic voice speaking your words.
• It is not free to use for meetings. See the note above.
• It does not replace a human interpreter where one is legally or contractually required.

Where to go next

The same setup covers Microsoft Teams and Google Meet, with their own device menus. For the general explanation of two-way meeting translation, start at the meetings hub. If a device is not showing up, or you are not sure which output Windows is actually using, work through the audio setup checklist — most Meeting-mode problems turn out to be one dropdown pointing at the wrong device. Known problems and their fixes live in troubleshooting.