Back to blog
Transcription Tips

Setting up MangoFinch with Zoom, Teams, and Google Meet

Step-by-step guide to using MangoFinch for real-time multilingual transcription alongside Zoom, Microsoft Teams, and Google Meet.

MangoFinch Team5 min read

MangoFinch does not replace your video platform. It runs alongside it — you stay in Zoom, Teams, or Meet exactly as you normally would, and MangoFinch captures the audio through your browser to provide real-time multilingual transcription.

This guide covers how to set that up for each platform, the technical reasons behind our browser-based approach, and the most common issues people hit during setup.

How MangoFinch connects to your meetings

MangoFinch runs entirely in your browser. When you open a session, it requests access to your microphone and your system audio (the sound coming out of your speakers or headphones). That second part is what captures the other participants in the call.

There is no plugin to install, no desktop app to download, and no bot that joins your meeting. Your video call stays exactly as it was. MangoFinch simply listens to the same audio you are already hearing and speaking.

The audio streams go to our speech engine over a WebSocket connection for transcription. Language detection happens per segment, translations get generated for any configured target languages, and everything renders in the MangoFinch interface in real time. Your video call does not know MangoFinch exists.

Why browser-based instead of API integration

We get this question a lot. Zoom, Teams, and Meet all have APIs. Why not integrate directly?

Three reasons.

Platform API limitations are severe. Zoom's real-time audio API requires an on-premise server deployment and a paid add-on license. Teams' Graph API provides post-meeting transcripts but not real-time audio streams. Google Meet has no public API for audio access at all as of early 2026.

Bot-based approaches create friction. Some transcription services join your meeting as a participant — you see "AI Notetaker" in the participant list. This works, but it makes people uncomfortable. In our user interviews, 8 out of 12 enterprise teams said they had either banned bot participants or had policies restricting them. A browser-based approach is invisible to other participants.

Browser audio capture is universal. WebRTC and the Web Audio API work the same way regardless of which video platform you are using. We wrote the audio capture code once, and it works with Zoom, Teams, Meet, Webex, and any other browser-based meeting tool.

The tradeoff: users need to join their meeting through a browser (or share system audio), and the setup takes about 60 seconds more than a native integration would.

Setting up with Zoom

Open your Zoom meeting in a browser tab. You can do this by clicking the meeting link and selecting "Join from Your Browser" instead of opening the Zoom desktop app. If you are already in the desktop app, that works too — you will just need to share system audio in step 3.

Open MangoFinch in a second browser tab. Go to app.mangofinch.com and start a new session. Select your source language and target languages.

When MangoFinch requests audio permissions, you will see two prompts. The first asks for microphone access. The second asks to share system audio — this captures what other participants are saying through Zoom.

On Chrome, the system audio prompt shows a "Share system audio" checkbox at the bottom of the screen-share dialog. Make sure this checkbox is selected.

Start talking. You should see transcription appear within 1-2 seconds.

Zoom-specific note: Zoom's browser client occasionally re-requests microphone permission after a network hiccup. If transcription stops mid-meeting, check that your browser still has microphone access.

Setting up with Microsoft Teams

Join your Teams meeting through the browser at teams.microsoft.com. Teams' browser client has full audio and video support on Chrome and Edge.

Open MangoFinch in a separate tab, configure your languages, and start a session.

Grant both microphone and system audio permissions when prompted. The process is identical to the Zoom setup.

Verify that both your voice and remote participants' voices are being captured. The MangoFinch audio level indicator shows two meters — one for your microphone, one for system audio.

Teams-specific note: Teams' browser client on Firefox has intermittent issues with system audio sharing. We recommend Chrome or Edge for Teams meetings. Also, Teams sometimes applies its own noise suppression aggressively, which can clip the beginning of words. If you notice transcription dropping the first syllable of sentences, go to Teams Settings > Devices > Noise suppression and set it to "Low."

Setting up with Google Meet

Join your Google Meet call as you normally would — it already runs in the browser.

Open MangoFinch in a second tab. Since both Meet and MangoFinch are browser tabs, Chrome manages the audio routing automatically.

Grant audio permissions. Meet is the simplest setup because there is no desktop app to complicate things.

One thing to watch: Google Meet's "Noise cancellation" feature can interfere with language detection for tonal languages like Mandarin and Vietnamese. The cancellation algorithm sometimes flattens tonal variations it interprets as background noise. If you are transcribing tonal languages, turn noise cancellation off in Meet.

Common issues and fixes

"I can hear everyone but MangoFinch is not transcribing remote participants." This almost always means system audio sharing was not enabled. When Chrome shows the screen-share dialog, look for the "Share system audio" checkbox. It is easy to miss — it is at the bottom of the dialog and unchecked by default.

"Transcription works but there is an echo — my words appear twice." Your system audio capture is picking up your own voice through your speakers. Use headphones instead of speakers.

"Transcription stops after 10-15 minutes." Check your browser's permissions. Some browser security settings revoke audio permissions after a timeout, especially in incognito mode. Use a regular browser window.

"The transcription is laggy — words appear 2-3 seconds late." Network latency between your browser and our transcription servers. This usually happens on VPN connections. If possible, exclude mangofinch.com from your VPN tunnel.

Tips for the best experience

Close unnecessary browser tabs. Each tab with active media adds audio to the system capture. MangoFinch will try to transcribe your background music along with your meeting.

Use a dedicated browser profile. Some teams create a Chrome profile specifically for MangoFinch sessions. This keeps permissions and extensions isolated.

Pin the MangoFinch tab. Pinned tabs do not get accidentally closed and they take up less space in the tab bar.

Test before important meetings. Run a 30-second test session before your actual meeting starts. The session persists — you do not need to restart it.

The setup takes about a minute the first time and about 15 seconds once you have done it before. The permissions persist across sessions.

Try MangoFinch free

Real-time transcription and translation for multilingual teams. No credit card required.

Start a free meeting