Google Meet Transcription Without Bots: 4 Ways Compared

You can get a Google Meet transcript without a bot in the attendee list in three ways. You can turn on Meet’s own transcripts (paid Workspace plans only), use a browser extension that reads Meet’s captions, or record the call audio on your own computer and transcribe it locally with Whisper. Only the local route keeps both the audio and the text on hardware you control. The other two avoid the bot but still put your transcript on someone’s servers.

For meetings where people sit in the same room, a fourth option is simpler than any of them: a phone on the table running an offline transcription app.

Why do people want Meet transcripts without a bot? #

Note-taking bots from Otter, Fireflies and similar services join the call as a named participant. Guests notice them, some hosts remove them on sight, and plenty of company policies block them outright. The bot also sends the full meeting audio to a third party, which is the part that matters if the call covers salaries, legal strategy, patients or unreleased products.

“Without a bot” and “private” aren’t the same thing, though. The table below shows where each option puts your audio.

Option 1: Google Meet’s built-in transcripts #

Google’s own transcription is the least work if your organization pays for the right plan. According to Google’s Meet help page, transcripts are available on Business Standard, Business Plus, Enterprise Starter, Enterprise Standard, Enterprise Plus, Teaching and Learning Upgrade, Education Plus and Workspace Individual. Free personal Gmail accounts don’t get them.

What to expect:

• The transcript is saved as a Google Doc in the organizer’s Drive, in a Google Meet folder.
• Everyone in the call sees a Transcripts icon once it starts, so it’s not secret.
• Supported languages are English, French, German, Italian, Japanese, Korean, Portuguese and Spanish.
• You can start a transcript from a computer or an Android device, but not from the iOS app.

Privacy-wise, this is Google processing Google’s own call. If your company already runs on Workspace and its admins are comfortable with Drive, that’s usually acceptable. If the whole point is keeping the words off cloud storage, it isn’t.

Option 2: Caption-reading browser extensions #

Extensions such as Tactiq don’t join as a participant. They run in Chrome, read the live captions Meet already generates, and turn them into a transcript with speaker names. Tactiq’s free plan covers 10 transcripts a month, and paid plans start at $8 per user per month on annual billing.

Two caveats. First, captions are produced by Google, so the audio is still processed in Google’s cloud. Second, the finished transcript lives in the extension vendor’s service, alongside any AI summary you request. Read the vendor’s data retention terms before you use one for sensitive calls. It’s a good fit for routine meetings where the only complaint was the visible bot.

Option 3: Record the audio and run Whisper on your laptop #

This is the only fully local method. You capture Meet’s audio on your computer, then feed it to an open-source Whisper engine such as whisper.cpp, which runs on Mac, Windows and Linux with no internet connection after the model download.

Step 1: Capture the call audio #

The catch nobody mentions: capturing your speakers records everyone except you. You need your microphone in the mix as well.

• Mac: install the free BlackHole virtual audio driver. In Audio MIDI Setup, create a Multi-Output Device (your speakers plus BlackHole) so you can still hear the call, and an Aggregate Device (your mic plus BlackHole) to record from.
• Windows: Audacity can record your speakers through the Windows WASAPI loopback input. Record your mic on a second track, or use a virtual cable such as VB-Cable.
• Linux: every output has a monitor source. Run pactl list short sources to find it, then record it with ffmpeg -f pulse -i <monitor-name> -ac 1 -ar 16000 meeting.wav. This works on PipeWire systems too.

Step 2: Transcribe after the meeting #

Whisper expects 16 kHz mono audio, so convert first, then run the command-line tool:

ffmpeg -i meeting.m4a -ar 16000 -ac 1 -c:a pcm_s16le meeting.wav
./build/bin/whisper-cli -m models/ggml-large-v3-turbo.bin -f meeting.wav -otxt -osrt

That writes a plain text file and an .srt file with timestamps. On a recent laptop, the turbo model handles an hour of audio in a fraction of an hour. On an older machine, use ggml-base.en.bin or ggml-small.bin and accept a few more errors.

Step 3 (optional): Live text during the call #

whisper.cpp ships a whisper-stream tool that transcribes a microphone or virtual device every half second:

./build/bin/whisper-stream -m ./models/ggml-base.en.bin -t 8 --step 500 --length 5000

Pick the capture device with -c. Live output runs a few seconds behind and makes more mistakes than the after-meeting pass, so many people record the file anyway and re-run it with a larger model.

What about speaker labels? #

Meet’s transcript and caption extensions know who is talking because Meet tells them. Local Whisper hears one mixed stream and doesn’t. If you need names, pyannote.audio adds speaker turns on your own machine. Its free community pipeline runs locally but needs a one-time Hugging Face token to download the model. Expect to fix labels by hand when people interrupt each other.

Option 4: A phone on the table for in-room and hybrid meetings #

When half the team is in a conference room, the simplest private transcript comes from a phone lying between them. Private Transcribe records on your iPhone or Android phone and runs Whisper on the phone’s own processor, so the audio never leaves the device and the app works in airplane mode after a one-time model download. It keeps recording when the screen locks, adds timestamps to every line, and lets you save names and jargon as custom vocabulary so they’re spelled right.

It also solves the “I recorded on my laptop but don’t want to set up whisper.cpp” problem. Move the recording to your phone by cable or AirDrop, tap Import, and the app transcribes audio or video files (m4a, mp3, wav, aac, mp4, mov) of up to 90 minutes on the phone. Be aware of its limits. It won’t label speakers, and it transcribes after you stop recording, so you won’t get live captions during the call.

For a cloud-free workflow across Zoom, Teams and in-person calls, see how to transcribe confidential meetings without cloud storage.

Tell people before you transcribe #

Meet shows an icon when its own transcription runs. Local recording shows nothing, so the notice is on you. Recording and consent laws differ by country and, in the US, by state. Some require every participant’s consent before you record. A one-line message at the start (“I’m recording this locally for my notes, and it won’t be uploaded anywhere”) covers most situations and tends to reassure people.

Getting accurate text from Meet audio #

Meet compresses audio heavily, and laptop mics pick up keyboards and fans. A few habits help more than switching models:

• Ask people to mute when not speaking. Overlapping voices cause most errors in every engine.
• Use a headset mic for your own voice so it doesn’t compete with the speaker output.
• Give Whisper your jargon. whisper.cpp accepts a --prompt with names and terms, and phone apps with a custom vocabulary setting do the same thing.
• Pick a larger model for important calls. Turbo and medium models noticeably beat base on accents and crosstalk.

Our guide to reducing background noise before you transcribe covers room and mic fixes in more detail. The same approach works for private Zoom transcription.

Frequently asked questions #

Does Google Meet have free transcription? #

Not for personal accounts. Meet’s transcripts are limited to certain paid Workspace editions, including Business Standard and above, some Education editions and Workspace Individual. Live captions are available more widely, but they aren’t saved as a transcript unless you use an extension that records them.

Can I transcribe a Google Meet call on my iPhone? #

You can’t start Meet’s own transcription from the iOS app. On an iPhone you can join from a laptop and record the room audio with an offline app on the phone, or import a recording made on your computer into a phone transcription app afterward.

Do caption extensions like Tactiq record my audio? #

They read the captions Meet already produces rather than capturing audio themselves, which is why no bot appears. The captions come from Google’s servers, and the transcript is stored by the extension’s company, so check its retention and AI-training terms.

Is local Whisper as accurate as Meet’s transcripts? #

On clean audio, a large or turbo Whisper model is competitive with cloud transcription. Meet’s version has the advantage of knowing who is speaking. Local Whisper is at its weakest with heavy crosstalk and poor laptop mics, so good audio matters more than the choice of engine.

Is it legal to transcribe a meeting without telling anyone? #

It depends on where you and the other participants are. Some jurisdictions only need one party’s consent, others need everyone’s. If you’re unsure, announce it at the start of the call.