Google Meet Transcription API for Live and Final Text
A Google Meet transcription API sends a bot into the call and returns a speaker-labelled transcript, live while people speak or after the call. From $0.45 per hour all-in.
*Get started in minutes*
What is a Google Meet transcription API?
A Google Meet transcription API turns a live Meet call into text your code can use. One request, and the transcript arrives at your webhook.
It does two jobs: capture the audio out of the meeting, then convert it to text. The meeting transcription API covers all three platforms.
What you get from every Google Meet call
One request in. Speaker-labelled text out, live while the call runs or once it ends.
*Get started in minutes*
Where Google Meet speaker labels come from
Speaker identity comes from the audio channel, not from a model reading a mixed track.
*Get started in minutes*
Meet gives up to three speaker streams at once, each tied to the active speaker, which is exact through normal turn-taking.
An utterance carries a name from the channel it arrived on. The speaker diarization entry sets out the other model.
Choose a transcription engine for Google Meet
MeetStream ships its own engine and routes audio to a provider you choose. Each figure below is that provider's own published number.
| Engine | Published language support | Real-time | Best for |
|---|---|---|---|
| MeetStream engine | In-house, auto language detection | No, post-call | One vendor, one bill, no extra key |
| AssemblyAI | 99 languages on Universal-2 | Yes | Broadest published language list |
| Gemini | 97 languages on the Live API | Yes | Teams already building on Gemini |
| Deepgram | 10 languages in real-time multilingual mode | Yes | Low-latency streaming workloads |
| Sarvam | 11 Indic languages plus English | No, post-call | Indian languages and accents |
| JigsawStack | Auto-detect or 162 languages | No, post-call | Batch jobs on a Whisper stack |
On Meet you can also choose the meeting captions engine, and the bot reads Google's own captions with nothing to set up.
Which Google Meet rules affect transcription?
Meet's rules are mostly about admission and edition. Each one links to Google's own documentation.
| Google Meet rule | What it means for a bot |
|---|---|
| Native recording editions | Limited to that include it. Recording also stops after eight hours. |
| Bots and admission | With "Anyone can ask to join" unchecked, third-party bots are automatically denied access. |
| Access types | Open, Trusted and Restricted decide who has to knock before joining. |
| Consent | Admins can require Gemini notes consent, and anyone who declines is removed. |
| Admin guidance | An admin article titled "External apps are recording Meet meetings" lists blocking app domains. |
How a Google Meet bot starts transcribing
A guest bot knocks, and someone in the call admits it. Under host management, hosts and co-hosts are the people who see that request.
Meet admits and records in one step, so text follows immediately. Participant events tell your app where the bot is.
From a Google Meet link to your product
One request puts a bot in the call. Text streams while people speak, then a webhook says the final transcript is ready.
*Get started in minutes*
Transcribe a Google Meet meeting with one request
Post the meeting link and name an engine. The bot joins, transcribes, and fires a webhook when the call ends.
curl -X POST https://api.meetstream.ai/api/v1/bots/create_bot \
-H "Authorization: Token $MEETSTREAM_API_KEY" \
-d '{
"meeting_link": "https://meet.google.com/abc-defg-hij",
"bot_name": "Acme Notetaker",
"recording_config": { "transcript": { "provider": { "meetstream": {} } } },
"callback_url": "https://your-app.example.com/meetstream"
}'Can I get a Meet transcript without a bot?
Google returns transcripts of its own, on Google's terms. A bot sits above them.
*Get started in minutes*
Google Meet transcription depends on the Workspace edition and the admin policy, and programmatic access follows those same rules.
A bot captures the audio directly and can stream text while the call runs. The Google Meet bot API page sets both routes side by side.
Live transcripts and the final record
Both come off the same bot. The question is when you need the words.
The Google Meet integration page covers signed-in bots, which join as named Workspace users.
Google Meet transcription API pricing
Capture costs $0.35 per bot-hour and transcription adds $0.10, so a transcribed meeting is $0.45 per hour all-in.
The rate falls toward $0.25 per bot-hour at volume, and every account starts with $5 of free credit. The pricing page has the full breakdown.
The comparison hub sets MeetStream beside the other meeting bot vendors, on their own published rates.
Google Meet transcription limits and defaults
Transcripts are kept for thirty days by default, or written into your own S3 bucket from the first byte.
One integration for all three platforms
The same request, the same webhook, the same transcript. Only the link changes.
Google Meet transcripts from a link, on any account.
One request captures the audio and returns the words, live or after the call.
*Get started in minutes*
Got a question? We got the answer.
Common questions about transcribing Google Meet calls. Where labels come from, which engines run, and what it costs.
*Get started in minutes*
Can I transcribe a Google Meet call through an API?
Yes. One request puts a bot in the Meet call, captures the audio, and returns a speaker-labelled transcript to your webhook.
Does Google Meet transcription need a specific Workspace edition?
Google's own transcription follows the Workspace edition and the admin policy. A bot captures the audio as a participant, whatever edition the host is on.
How are Google Meet speakers labelled?
Meet gives up to three speaker streams at once, each tied to the active speaker, so an utterance carries a name through normal turn-taking.
Can I use Google Meet native captions?
Yes. Choose the meeting captions engine and the bot reads the captions Meet produces, with nothing to set up on the Google side.
Can I get a live Meet transcript during the call?
Yes. Choose a streaming engine and text chunks arrive at your webhook while people are still speaking, ready for live coaching or an in-meeting assistant.
Which transcription engines work on Google Meet?
The MeetStream engine plus AssemblyAI, Gemini, Deepgram, Sarvam and JigsawStack, or Meet's own captions. Pick one per request.
What does a Google Meet transcript contain?
A list of utterances, each tied to a named participant with start and end times, so a summary or a search index needs no extra speaker step.
How much does the Google Meet transcription API cost?
Capture is $0.35 per bot-hour and transcription adds $0.10, so $0.45 per hour all-in, falling toward $0.25 at volume. Every account starts with $5 of credit.
Transcribe your next Google Meet call
$5 of free credit, self-serve API keys, and no sales call between you and your first transcript.