Client calls
No bot to explain, and a follow-up drafted from what was actually said.
Mr.Speech is meeting transcription software that records both sides of a call on your own computer, labels who said what, and turns the transcript into AI meeting notes, action items and a task board. No bot joins the meeting, it works with Teams, Zoom and Google Meet, and you choose between a fully local mode and European cloud processing.
Most meeting transcription software sends a bot into the call. It appears in the participant list, has to be explained to clients, and is blocked in many corporate tenants. Mr.Speech takes the other route: it records on your computer. Your microphone and your system audio are captured as two separate tracks, so it works with Microsoft Teams, Zoom, Google Meet or any other conferencing tool, because it records the machine rather than the meeting service. An in-person mode records a single microphone for a room with no call at all, with optional live subtitles for the audience.
When you stop, the recording is transcribed either on your device or on European infrastructure, and an on-device speaker identification model labels who said what. Name a person once in a finished meeting and their voice is remembered locally for future meetings, without any voice data touching the cloud. One click then produces a structured protocol, a summary, action items or a follow-up email draft, and the action items land on a Kanban board with due dates and reminders. The AI is instructed never to invent a decision, a task or a name that was not in the transcript.
Teams meeting transcription without recording, honestly. Teams has its own transcription, and Microsoft's documentation says it can run as a separate switch from a Teams recording, that the organiser and the person starting it both need the transcription policy turned on by an admin, that a qualifying Microsoft 365 licence is needed to record, and that transcripts are stored in OneDrive and SharePoint. Mr.Speech records the audio on your computer instead, so nothing is switched on inside Teams, no bot appears and no admin policy is involved. That is not a way to record quietly. You are still recording, so participants must be informed, and in many places their consent is a legal requirement. Because Teams is not doing the recording, Teams will not announce it for you: say it yourself before you press the button. This page describes the product and is not legal advice.
Your microphone and your computer's audio are recorded separately, so you and everyone else are labelled correctly from the first second, whatever the conferencing tool.
An on-device speaker identification model recognises you and every other person speaking. Named voices are remembered locally for future meetings; no voice data touches the cloud.
The on-device mode keeps recording and transcription on your machine. The cloud mode is processed in Europe; the processors are named in our privacy policy.
One click for a protocol, a summary, action items or a follow-up draft. Action items become a Kanban board with due dates and reminders, in the same app.
There is no sentiment scoring, no engagement metric and no talk-time ranking presented as attention. It writes down what was said and who said it, nothing more.
Every cell is checkable on the vendor's own pages. Where a vendor does not state something, the cell says so rather than guessing. Otter and Fireflies both offer botless desktop recording too; the differences are elsewhere.
| Mr.Speech | Otter.ai | Fireflies.ai | Microsoft Teams transcription | Granola AI | |
|---|---|---|---|---|---|
| Records without a bot | Always, on your own machine | Desktop app records without a bot | Botless desktop recording per their guide | Built into the meeting, but only inside Teams | Uses the computer's audio, no bot invited |
| Zoom, Google Meet and in-person meetings | Any conferencing tool, plus an in-person mode with one microphone | Not stated by the vendor on the pages we checked | Not stated by the vendor on the pages we checked | Teams meetings and events only | Zoom, Google Meet, Teams and every other meeting app, per the vendor |
| Fully on-device transcription | Recording and transcript stay on your machine | Uploads and transcribes once a connection returns | Not documented either way | Microsoft cloud; transcripts stored in OneDrive and SharePoint | Not stated by the vendor; notes are stored in a US-hosted cloud |
| Where your data lives | Your machine, or European cloud with processors named in the privacy policy | Not stated by the vendor on the pages we checked | Not stated by the vendor on the pages we checked | OneDrive and SharePoint; region not stated on the page | US-hosted AWS cloud; audio deleted after transcription, per the vendor |
| Runs on Linux | Native, alongside macOS and Windows | Mac and Windows desktop only | Windows 10+ and macOS 12+ only | Not stated by the vendor on the page we checked | macOS, Windows, iOS and Android |
| Dictation into any app | Hotkey anywhere, cleaned text typed in | Meeting transcription only | Meeting transcription only | A meeting feature, not dictation | Meeting notes; no dictation described |
| Notes and task boards in the same app | Notebook, wiki links, semantic search, Kanban | Meeting notes and summaries | Searchable conversation archive | Transcript stored with the recording; no task board described | Notes, action items and follow-ups, per the vendor |
Otter.ai and Fireflies.ai checked on their own pages in July 2026; Microsoft Teams (learn.microsoft.com and support.microsoft.com) and Granola AI (granola.ai) on 26 September 2026. Products change, so verify anything decisive yourself.
From the start of the recording to filed notes in six steps.
Live transcription to follow along as people speak, recording with cloud transcription for maximum accuracy processed in Europe, or recording with on-device transcription for meetings that must never touch the cloud. You decide per meeting.
Say that you are recording, and what for, before you press the button. Participants must be informed; in many places that is the law, and it is good practice everywhere. Teams will not show its own recording notice, because Teams is not doing the recording. This page is not legal advice.
Microphone and system audio are captured as separate tracks. If your mic is muted or system audio is silent, the app tells you within seconds. Audio is written to disk continuously, so a crash does not lose the meeting. If you connect Google Calendar read-only, the app spots when a call starts and offers to record.
Speaker labels, a structured protocol, a summary, action items and a follow-up email draft. Action items go straight onto a task board with due dates and reminders.
Export as text, Markdown, SRT or JSON, or file the meeting as a note with wiki links and search by meaning.
If your past meetings live in Granola or Wispr Flow, Mr.Speech imports transcripts, summaries, attendees and folders into their own project, so the archive does not start from zero.
No bot to explain, and a follow-up drafted from what was actually said.
A single microphone captures the room, with optional live subtitles for the audience.
The on-device mode keeps both the recording and the transcription on your machine.
Yes. Mr.Speech records your computer's audio and your microphone, so it works with Microsoft Teams, Zoom, Google Meet and any other tool, including tenants that block bots. Teams also has its own transcription, which Microsoft says needs the transcription policy turned on for the organiser and the user and stores transcripts in OneDrive and SharePoint.
Teams' own transcription is a separate switch from a Teams recording, according to Microsoft. Mr.Speech records on your computer instead, so nothing is switched on inside Teams at all. You are still recording, though: tell the participants before you start, because Teams will not announce it for you. This is not legal advice.
It depends on whether you accept a bot, where the audio may go and what happens after the transcript. Fireflies is convenient when a bot that auto-joins calendar meetings means nothing is missed. Otter has years of team workflow. Granola AI records from the computer's audio and stores notes in a US-hosted cloud, per its own site. Teams transcription is the simplest if every meeting is in Teams and your admin allows it. Mr.Speech is the choice when no bot may appear, when the transcript must be able to stay on your machine or in Europe, and when the notes should become tasks in the same app.
That depends on your setup, not on a badge. Mr.Speech gives you the building blocks: an on-device mode where audio never leaves the machine, a European cloud option, processors named in the privacy policy, and no emotion or sentiment analysis. You must still inform participants and document your own use. This is not legal advice.
No. Recording happens on your computer, so other participants see no extra attendee, and it keeps working where bots are blocked.
Yes. The on-device meeting mode transcribes entirely on your machine when you stop. Nothing is uploaded.
The cloud mode covers around 100 languages, detected automatically. The on-device engine supports 40 languages with real punctuation.
Yes, as text, Markdown, SRT or JSON, or as a note in the built-in notebook, which exports as plain Markdown.
Yes. The in-person mode records a single microphone for a room with no call at all, with an optional presentation captions view that shows live subtitles to the audience.
From Granola and Wispr Flow, yes: transcripts, summaries, attendees and folders come over into their own project. Other sources are not claimed today.
Mr.Speech is meeting transcription software that records both sides of a call on your own computer, labels who said what, and turns the transcript into AI meeting notes, action items and a task board. No bot joins the meeting, it works with Teams, Zoom and Google Meet, and you choose between a fully local mode and European cloud processing.