Monday, 10 a.m., a call with a client. Before anyone has even said hello, a new tile appears in the participant list: "Jake's Notetaker". The client asks who will keep the recording, and whether they can say no. To transcribe a Teams meeting without a bot, you need a different method: send nothing into the call, and transcribe on your PC the sound it's already playing, your microphone included. The same method works for Zoom, Meet or a webinar, without any organizer rights.
The short version
- A note-taking bot joins the call like a guest and sends the audio to a service's servers. A transcription done on your PC adds no one to the meeting.
- It works with any app that plays sound, even when you're not the organizer.
- To get the name of whoever is speaking, you need software that separates voices, then lets you name them.
- With Ondula, the live meeting is transcribed offline, on the PC, with speakers identified: that's Lifetime Local, €19.99 once. The PDF meeting summary comes with Cloud.
- No bot doesn't mean in secret: always tell the participants.
Why note-taking bots bother people so much
You give a service the meeting link, and an automated account joins on your behalf, listens, transcribes on its own servers and sends you the text. Convenient for whoever invited it, much less so for everyone else.
- External participants didn't choose it. A client or a candidate sees a stranger enter the call, and speaks less freely.
- The organizer often has to admit it from the waiting room. If they refuse, your meeting notes go with it.
- Some companies block it in their video conferencing settings, or ban it in their contracts.
- The audio goes to a third party. For an HR interview, a board meeting or a conversation under an NDA, that's often where it ends. On October 20, 2026, Google's suggestions for Otter.ai, one of the best-known services, included "privacy concerns" and "privacy lawsuit".
Bot or PC sound: what's the difference?
Without a bot, the software listens to what your computer is already playing, in other words what comes out of your headphones. The video conferencing app sees nothing.
| Bot in the meeting | Transcription on the PC | |
|---|---|---|
| Visible to participants | Yes, like a guest | No: nothing joins the call |
| Organizer's approval | Often (waiting room) | No |
| Supported apps | The ones the service can join | Any app that plays sound |
| Where the audio goes | To the service's servers | It stays on the PC (local) |
| A meeting you're not in | Possible | Impossible: the PC has to be in the call |
What about Teams' built-in transcription?
Teams has its own live captions and transcription. No bot, but everything is decided on the meeting's side. Translated captions require Teams Premium or a Copilot license (Microsoft Support, checked on 10/20/2026). And according to Bluedot, a meeting tools vendor, you can't get a Teams transcript without starting the meeting recording, which everyone sees and the organizer controls.
As for recognizing voices, Microsoft only offers it, according to our check on 10/20/2026, in Teams Rooms, equipped meeting rooms, with voice profiles stored in the organization's cloud. From your laptop, as a simple guest, or on Zoom and Meet, you depend on what others have enabled. For live translation outside Teams, we compared the options in our article on translated live captions on Windows.
How do you transcribe a Teams meeting without a bot?
The method works with any software that captures your computer's sound.
- Tell the participants, right at the start: "I'm transcribing for the notes; the text stays on my computer." One sentence is enough, and it's often a legal requirement.
- Choose software that captures both the computer's sound and your microphone. The computer's sound alone doesn't contain your voice, since Teams doesn't play back what you say. The microphone alone only hears you.
- Wear a headset or earbuds. With speakers, the meeting's sound comes back into your mic: sentences get doubled and speakers get mixed up.
- Start the transcription before the call. The first two minutes are often when the agenda gets set.
- Name the voices early. "Speaker 2" becomes "Claire, purchasing" from her first sentence, and the rest reads itself.
- Proofread before sharing. One mangled proper name or a sentence attributed to the wrong colleague is enough to get meeting notes challenged.
Two traps you rarely read about. Mute Windows and chat app notification sounds during the call, or they'll end up in the text. And don't play a video in another tab: everything the computer plays gets captured.
How do you know who's speaking in the transcript?
To attribute each sentence, software has to do two distinct things. First, separate the voices: detect that at a given moment a different person is speaking, and number each voice. Then, recognize the voices: keep a kind of audio signature for each one, so it can find the same person next week without renaming everything.
The first step is common. The second is much rarer on Windows without going through a server: when we searched on October 20, 2026, we mostly found Mac tools and a developer's project on GitHub.
One detail changes everything: your own voice. If the software listens to your mic separately, it knows for certain when it's you speaking. The limit, finally, is the same for everyone: two similar voices can be confused, and when everyone talks at once, attribution gets approximate.
What Ondula does during a meeting
Ondula is a voice dictation app for Windows 10 and 11. Its Meeting feature writes out, as it happens, everything your PC says, in a panel: Teams, Zoom, Meet, Discord, a browser or a broadcast.
- No bot. Ondula listens to the computer's sound. Nothing joins the call, and nobody needs to approve it.
- Your mic, when a call is using it. As soon as another app uses your microphone, Ondula adds it to the computer's sound, and your sentences appear under "Me". The panel tells you where things stand: mic captured, not captured because no app is using it, or blocked if you've forbidden it.
- Each turn under the name of whoever is speaking. Click a name to change it. Two people mixed up can be split with "Separate the voices", one person counted twice can be grouped with "Merge with", and an off-topic passage can be deleted.
- Voice memory. A voice named once is recognized from one meeting to the next. You're the one who names people: Ondula doesn't guess any names.
- Offline, on the PC. The live meeting runs locally, with the same model as dictation (NVIDIA's Canary): 25 European languages, with the spoken language detected automatically. It uses the graphics card, whatever its brand, or else the processor.
- Transcribed or translated. The panel shows the original language or a translation: from a language into English, or from English into any of the 24 other languages. Between two languages other than English, German to French for example, there's no direct translation.
- The meeting summary. The transcript goes into your history. With Cloud, it becomes a PDF summary: participants, summary, decisions, action items. For that summary, it's the text that goes to the server, not the audio.
The app says so itself: without a graphics card, speaker recognition can be less reliable. Run a test on your own laptop before a meeting that matters.
The steps with Ondula
- Install Ondula from the Microsoft Store.
- Activate Lifetime Local (€19.99, once), or get Cloud, which includes it.
- In Settings, Capture tab, assign a shortcut to the Live meeting card: it has none by default. You can also add "Meeting" to the palette, in the Personalization tab.
- Put on a headset and join the meeting as usual.
- Press your Meeting shortcut. The panel opens on the right side of the screen; check that it shows your mic is being captured.
- Name the speakers as you go, then reread the transcript in your history after the call.
For a meeting you've already recorded, or an interview, use file transcription instead, which is free: see our guide to transcribing audio or video to text. And if you're leaving an online service, our Otter.ai alternatives article covers the switch.
How much does meeting transcription without a bot cost?
- Free, no account: unlimited offline dictation, transcribed files, history.
- Lifetime Local: €19.99 once, no subscription, on 3 PCs. It adds live meetings and subtitles, speaker recognition and voice memory.
- Cloud: €5.99 a month, cancel anytime. Everything in Lifetime Local, plus 20 hours of online audio and 20 hours of summaries every month. Cloud Pro: €9.99 a month, 40 hours and 40 hours, for back-to-back meetings.
- Trial: 14 days of Cloud free when you create an account, with 10 hours online, no credit card.
No per-participant or per-seat billing. Details are on the pricing page.
The limits, plainly
- Windows only: no Mac, no phone.
- The PC has to be in the call. That's the flip side of having no bot.
- A dictation pauses the meeting transcript while you speak. During an important meeting, take your own notes on the keyboard.
- You capture what you hear. A participant who sounds bad on your end will be transcribed badly, and overlapping voices remain hard for every tool.
Frequently asked questions
Can participants see that I'm transcribing?
No: no participant is added and the video conferencing app sees nothing. That's exactly why you need to tell them yourself, before you start.
Do I need to be the Teams meeting organizer?
No. The software listens to your computer's sound and needs no rights over the meeting, on Teams or anywhere else.
Is my own voice in the transcript?
With Ondula, yes, as soon as a call uses your mic, unless you forbid it. Your sentences appear under "Me".
Does the meeting go over the internet?
No: the live meeting is transcribed on your PC and the audio never leaves it. Only Cloud's summary sends the text online. Our article on voice typing, GDPR and privacy explains the difference; other answers are in the FAQ.
Try it without a bot at your next meeting
Start for free with a recording you already have. If you like the result, live meetings cost €19.99, once. Install Ondula before your next call, put on a headset, and nobody will see an unexpected guest walk in.
