Meeting Transcription for Google Meet, Zoom and Teams (2026)

Meeting transcription means getting a written record of a Google Meet, Zoom or Microsoft Teams call, either live while the meeting runs or afterwards from a recording. You have three practical routes: switch on the platform's built-in transcript, run a browser extension that listens to the meeting tab, or record the call and upload the file to a transcription service. This guide compares them, summarises the consent rules that apply in the EU, and shows how to go from a raw transcript to usable minutes.

Why transcribe meetings at all

The obvious benefit is that nobody has to take notes, so everyone can actually participate. The less obvious ones matter more over time:

• Decisions and action items stop getting lost. A searchable record settles "who agreed to what" without a second meeting.
• People who missed the call catch up in five minutes with a summary instead of a 60-minute recording.
• Colleagues working in a second language can read at their own pace or translate the text.
• Accessibility: live captions and transcripts make meetings workable for deaf and hard-of-hearing participants.
• Interviews, user research and sales calls become data you can analyse rather than memories you reconstruct.

The flip side is that a transcript is a permanent record of what people said, which is exactly why the legal section below exists.

Consent and recording laws in the EU, briefly

This is not legal advice, but the general picture for the EU is consistent enough to summarise.

GDPR applies. A recording or transcript of a meeting contains personal data: voices, names, opinions, sometimes health or HR details. Processing it requires a legal basis (usually legitimate interest or consent), a clear purpose, a limited retention period and transparency towards the participants. For companies, the transcription provider becomes a data processor, so a data processing agreement and a check on where the data is stored are part of due diligence; our article on GDPR and transcription goes into the details.

National laws add consent requirements. Many member states, including Germany (§ 201 StGB), France, Italy, Austria and others, make it a criminal offence to record a non-public conversation without the consent of the participants. The practical consequence is that "all-party consent" is the safe default across the EU, regardless of which country the participants are in.

Workplace rules apply on top. Works councils in Germany and employee representation bodies elsewhere often have a say in recording tools; many companies have an internal policy on which tools are permitted.

What this means in practice:

1. Announce recording or transcription at the start, and repeat it for late joiners.
2. Use the platform's built-in notification where available (Meet, Zoom and Teams all show a banner when native recording or transcription starts).
3. Offer an alternative for people who object, such as turning transcription off or leaving the sensitive part untranscribed.
4. Decide up front how long you keep the recording and who can access it.
5. Pick a provider whose data location and retention policy you can explain to a participant who asks.

Option 1: Built-in transcription in Meet, Zoom and Teams

All three major platforms now offer transcription, with differences in who can turn it on and what you get.

Google Meet

Google Meet offers live captions for free in several languages, and a Transcripts feature on Google Workspace plans that saves a transcript to Google Docs in the organiser's Drive after the meeting. Recording is also available on eligible Workspace editions. Participants see a notification when transcription starts. Google's help page at https://support.google.com/meet/answer/12849897 lists the eligible plans and supported languages.

Zoom

Zoom provides live captions and audio transcript generation for cloud recordings on paid plans; the transcript is saved alongside the recording and can be downloaded as VTT. Zoom's AI Companion adds meeting summaries. Local recordings do not get an automatic transcript, so you upload the file elsewhere if you need one. Hosts control whether participants may record.

Microsoft Teams

Teams offers live transcription and recording on Microsoft 365 business and enterprise plans. The transcript is stored in OneDrive or SharePoint, shows speaker names taken from the Teams accounts (a real advantage over voice-based diarization), and can be downloaded as DOCX or VTT. Admins can restrict who may start transcription.

Built-in options are convenient and well integrated with your calendar and storage. Their limits are plan-dependence, uneven language support, basic editors, and the fact that data goes wherever the platform stores it, which for US-based vendors is a point to clarify in your GDPR assessment.

Option 2: A browser extension that listens to the meeting

A browser extension captures the audio of the meeting tab and sends it to a transcription service in real time or at the end of the call. The Audiotexta extension works this way with Meet, Zoom (web client) and Teams (web client), and competitors such as Otter and tl;dv offer similar extensions or bots that join the call.

Advantages:

• Works regardless of your plan on the meeting platform or whether you are the host.
• One transcript workflow across all three platforms, with the same editor, speaker labels and exports.
• No bot joining the call, which some participants find intrusive.
• Transcripts land in your workspace with timestamps, ready for summary and translation.

Limitations:

• You must use the web version of the meeting tool, not the desktop app.
• Speaker identification relies on voice rather than account names, so you may need to rename speakers afterwards.
• Legally it is identical to recording: participants must be informed, and the extension does not show a banner inside the meeting on your behalf.

Option 3: Record, then upload

The simplest and most portable approach: record the meeting (natively, with a screen recorder, or with a phone on the table for an in-person meeting), then upload the MP4, M4A or MP3 to a transcription tool. This also covers recordings you already have from the past year.

In Audiotexta, you drop the file into the web app, choose the language, enable speaker labels, and get a transcript with timestamps a few minutes later. The process is described step by step in how to transcribe audio to text. The disadvantage is the delay and the extra file to manage; the advantage is that it works with anything, including hybrid meetings where half the room is on one microphone.

Comparison table

From transcript to minutes

A raw transcript is a record, not minutes. Turning one into the other is where most of the value is, and it takes about ten minutes if you follow a routine.

Clean the speakers and the terms

Rename "Speaker 1" to actual names; with four or fewer participants this is quick, because voices are consistent. Fix product names, client names and acronyms once with find-and-replace.

Summarise with AI, then verify

Generate a summary with key points and action items. Good tools, Audiotexta's included, produce something close to a first draft of minutes. Read it against the transcript, because summaries occasionally merge two similar points or attribute a decision to the wrong person.

Add what the model cannot know

Owners and due dates for action items, links to the documents mentioned, and the context for anyone who was not present.

Share with the right scope

Minutes can go to the whole team; the full transcript often should not, especially if it contains frank discussion or personal details. Decide the retention period and delete the recording when it has served its purpose.

A minimal minutes template:

• Date, participants, purpose
• Decisions (one line each)
• Action items (owner, due date)
• Open questions
• Link to the transcript (restricted access)

Practical tips for better meeting transcripts

• Headsets beat laptop microphones. The biggest accuracy gain available is asking remote participants to use a headset.
• One person at a time. Crosstalk is the hardest problem for any speech recognition system; a good moderator helps the transcript as much as the meeting.
• Mute when not speaking to remove keyboard noise and background conversations.
• Introduce yourselves at the start. It gives you a reference for renaming speakers.
• Say numbers and names clearly, especially amounts, dates and project codes.
• Hybrid meetings need a room microphone. A laptop in the middle of a conference table produces the worst audio of any setup; a conference speakerphone is a cheap fix.
• For interviews and user research, our interview transcription guide covers consent forms and anonymisation as well.

Choosing a setup

If your organisation already pays for Workspace, Zoom or Microsoft 365 with transcription included, start with the built-in option and evaluate whether the output is good enough. If you move between platforms, are rarely the host, work in a language the platforms support poorly, or want one place for all transcripts with a proper editor, a browser extension or upload workflow makes more sense. And if data protection is part of the decision, which under the GDPR it often is, check the provider's privacy terms, retention settings and data processing agreement, as well as the meetings feature page, before you roll anything out to a team.