Do You Need a Deepgram API Key to Translate Google Meet?

A common misconception: you need a Deepgram API key to use MeetVoice. Here’s the truth: you don’t.

Speech recognition is included in every plan — and in the 30-minute trial, before you pay anything. The relay handles Deepgram integration on your behalf. You don’t need to set up an account, generate credentials, or paste anything into settings. Install the desktop app, install the Chrome extension, pick your language pair, and translation starts working. This is one of the biggest changes in MeetVoice 2.0: we removed the friction that used to block most new users before they ever heard a translated sentence — the full story is in the 2.0 announcement.

The default: speech recognition included via relay

With MeetVoice 2.0, we move away from the “bring your own key” model that required most users to jump through hoops with a third-party API provider. If you want the wider picture of how real-time translation in Google Meet actually works, that post covers the whole pipeline.

Here’s how it works now: your meeting audio flows from your microphone → through the MeetVoice desktop app → through the MeetVoice speech recognition relay (our Deepgram account) → back to you as transcripts. The relay is a pass-through: it authenticates your license, forwards the audio stream, counts the time against your subscription quota, and returns the transcript. Audio is never stored — it passes through in transit and is gone.

Your monthly speech recognition quota depends on your plan:

• Trial — 30 minutes total (no credit card, no commitment)
• Starter (€9/month) — 5 hours per month
• Pro (€29/month) — 20 hours per month
• Business (€79/month) — 80 hours per month

Most users fit comfortably within these limits and never think about it again. Full plan details and checkout are in the pricing section of the home page.

When you might want your own Deepgram key

A Deepgram API key is optional. You can add your own if you have a specific reason. Here are the scenarios where it makes sense:

A direct relationship with the speech vendor. Some organizations have vendor policies that favour contracting with the speech-recognition provider directly, with no intermediary in the audio path. If that’s you, go to Settings → Keys → Deepgram, paste your API key from console.deepgram.com, and audio goes from your computer straight to Deepgram over TLS — the MeetVoice relay is not in the audio path at all, and your contractual relationship is with Deepgram rather than with us. To be clear about what this is and isn’t: it removes a hop and changes who you have an agreement with. It is not end-to-end encryption, and it says nothing about where Deepgram processes or stores your data — so don’t treat it as a compliance guarantee for any particular regime.

High-volume transcription. If you already have a Deepgram account and run high-volume meetings with heavy translation usage, using your own key means your quota is separate from your MeetVoice plan quota. With your own key, Deepgram bills you for your actual usage, and the included monthly time in your MeetVoice plan no longer applies. However, for most users, the included quota in a MeetVoice plan is cheaper and simpler — you don’t have to manage Deepgram billing separately.

Existing API integration. If you already subscribe to Deepgram for other transcription work (customer support recordings, podcast transcription, etc.) and manage all APIs through a single vendor account, using your key keeps billing and support in one place.

Usage visibility in Deepgram’s own dashboard. Some teams would rather track their transcription minutes where the rest of their Deepgram usage already lives, in the vendor’s native console. Adding your own key puts the numbers there.

The key field is one-time setup — add it once and it stays active until you remove it. You can switch back to the relay at any time by deleting the key.

You probably don’t need to worry about this

Unless you specifically need direct API access or have a Deepgram account you prefer to use, the default relay just works. It’s covered by your plan — and during the 30-minute trial, before you pay for one. You get the same speech recognition quality as a direct Deepgram integration, no extra complexity, and no configuration friction.

The beauty of the relay model is that it solves the most common objection to MeetVoice: “Why should I set up yet another third-party account?” Now the answer is simple: you don’t have to. We handle it.

Getting started with MeetVoice

If you’re new to MeetVoice, here’s all you need to do:

1. Download the desktop app from the download page (Windows or macOS)
2. Add the MeetVoice Chrome extension from the Chrome Web Store
3. Start the free 30-minute trial — no credit card required
4. On your first Google Meet call, pick your language pair and click “Translate”

That’s it. The speech recognition relay and translation are already running. The quick setup guide walks you through the language selection step if you need extra guidance.

Your speech recognition quota will reset automatically at the start of each calendar month. As long as you stay within the limits of your plan, you’ll never see a surprise bill or API error. The quota is baked into your plan — it’s one less thing to manage.

Questions?

If you have questions about your plan tier, how much quota you’re using, or how to add your own Deepgram key, write to [email protected] — we’re happy to help. Most questions can be answered in a couple of messages.