New in 5.0 Dictation in any app

Meetings, debriefed.
Everything else, dictated.

Private AI for everything you say. Record, transcribe, and summarize meetings, and dictate into any app. It runs on your computer, and you pay for it once.

No account, no credit card. macOS and Windows.
  • No bot joins your call
  • Nothing is uploaded, ever
  • Works offline after setup
  • One payment, no subscription
New in version 5.0

Talk instead of type. In any app.

Hold the shortcut anywhere on your computer, say what you want written, and let go. MeetingDebrief transcribes it on your device and types it into the app you were already in. It uses the same local speech engine as your meetings, so nothing leaves your computer.

Try it: pick an app and hold the button.

Listening…

esc Discard

Press and hold the button, or hold on your keyboard. Let go to paste.

Works in any app

Editors, terminals, chat, email, browser forms. If it has a text cursor, dictation can type into it.

Typed out as you let go

Words appear in the dictation window while you speak. A 15-second dictation is pasted within a fraction of a second of releasing the key.

Your clipboard survives

Copy a log, dictate "take a look at this error", and the log is still on your clipboard to paste underneath.

Music pauses itself

Other audio is muted while you speak and restored exactly as it was afterwards, volume included.

Spells your jargon

Dictation uses the same Vocabulary as your meetings, so product names and colleagues come out spelled right.

A history you can search

Every dictation is saved with its audio. Dictated into the wrong window? Copy it again from the tray.

Take the app for a spin.

This is MeetingDebrief 5.0. Open this page on a larger screen to click through an interactive replica of the app.

From the first word of the meeting to the last follow-up, MeetingDebrief handles every step with AI that runs on your own hardware.

01

Record

Hit record and MeetingDebrief captures your microphone and the other side of the call, on any meeting app. Echo cancellation and noise suppression clean it up, and live captions show what it hears. No bot ever joins your call. Already have a recording? Drop in an MP3, WAV, or M4A.

02

Transcribe

Parakeet or Whisper turns speech into punctuated, time-stamped text on your own GPU, labelled Me and Other. It runs in the background, so the app stays usable. Wrong language detected? Re-transcribe in the right one with two clicks.

03

Summarize

Pick a template and get the decisions, action items, owners, and deadlines. Summaries know the date of the meeting, so "by Friday" means the right Friday, and they flag themselves when the transcript changes.

04

Ask

Chat with any meeting. "What did Sarah commit to?" gets an answer with a timestamp you can click to hear the moment it came from.

05

Find

Search every transcript, summary, note, and dictation at once. Organize with folders and tags, add your own notes, and mirror everything to Markdown files for Obsidian or Notion.

Ask the meeting, not your memory.

Every answer comes from the transcript and your own notes, with a timestamp you can click to hear the moment it came from. Hours-long meetings stay answerable, and the whole conversation happens on your machine.

Teach it your words once.

Names, acronyms, and product jargon are what speech models mangle. Vocabulary primes the transcriber, corrects the transcript as it is saved, and explains your terms to the model that writes summaries. Fix a word once and MeetingDebrief offers to remember it, or let Suggest terms pull the jargon out of a meeting for you. Dictation uses the same list.

Plays well with your AI tools.

Dictate your prompts

Describe the bug to Claude Code or Cursor out loud. Dictation types into terminals and editors, and your Vocabulary keeps framework names intact.

Built-in MCP server

Give Claude, Claude Code, and Cursor read-only access to your meetings. Paste one config and your agent can search transcripts and pull up what you agreed last week.

Markdown and a local API

Session Mirror writes every transcript and summary to Markdown files on disk. A local HTTP API with Swagger docs covers the rest.

Your voice. Your data. Period.

MeetingDebrief is local-first by design. Every recording, transcript, summary, and dictation lives only on your device. We don't run servers that process your audio. We don't collect analytics. We have no way to access your meetings, because they never reach us. When we say private, we mean the architecture, not a paragraph in a policy.

No monthly fee to use your own computer.

Meeting notes and dictation usually mean two subscriptions and two companies holding your recordings. MeetingDebrief does both with your processor and your memory, so there is no server bill to pass on. One purchase. Yours forever.

New in version 5.0

  • Dictation

    Hold a shortcut anywhere, speak, and the text is typed into the app you are using.

  • Parakeet, recommended

    The fast engine leaves preview. It stays loaded, so transcription and dictation start instantly.

  • Dark mode

    Every window follows your system appearance, or pin it to light or dark.

  • Date-aware summaries

    "Next quarter" and "by Friday" resolve against the day of the meeting.

  • Pick your speakers on Windows

    Choose which output device to capture, so headset calls are recorded correctly.

What people are saying.

“The Markdown mirror is what sold me. My project notes are in Obsidian already; I really didn't want another place to remember to search.”

Felix Weber

Backend developer, freelance

“Had a question about whether we'd agreed to remove something or just deprecate it. The chat pointed me to the right part of the discussion and I listened back. That's the sort of thing I want this for.”

Marcus Reed

Engineering manager, payments

“No card for the trial. I installed it, used it for a few planning calls and decided from there. Appreciated being able to do that without setting a cancellation reminder.”

Aisha Rahman

Operations lead, online education

FAQ

Is MeetingDebrief really completely private?
Yes. Recording, transcription, summaries, chat, and dictation all run on your device. No audio or text is sent to any server, ever. We don't collect analytics, telemetry, or usage data of any kind.
How does dictation work?
Hold the dictation shortcut (Control+Alt+D by default) anywhere on your computer, say what you want written, and let go. MeetingDebrief transcribes it on your device and types it into the app you were using: an editor, a terminal, a chat box, an email. You can also tap once to start and again to stop. Every dictation is saved in a searchable history.
Is dictation included in the price?
Yes. Meeting notes and dictation are one app and one license. There is no add-on to buy and no usage limit.
I already own MeetingDebrief. Do I get 5.0 and dictation?
Yes, free. Every update is included with your license, major versions too. Update from inside the app.
Does it work with virtual meetings?
Yes. MeetingDebrief captures your microphone and your computer’s audio output, so it works with Zoom, Google Meet, Microsoft Teams, Slack huddles, or any other app. No bot joins the call. It also works for in-person meetings using just the microphone, and you can import existing recordings.
How accurate is the transcription?
MeetingDebrief ships two local speech engines. Parakeet is the recommended engine: fast, punctuated, and available in 25 languages. Whisper covers 99+ languages for meetings. Accuracy depends on audio quality, so the app also cleans up the audio (echo cancellation, noise suppression) and lets you teach it names and jargon through Vocabulary.
What are the system requirements?
On macOS: macOS 13 (Ventura) or later, with Apple Silicon (M1 or newer) recommended for optimal performance. Intel Macs are supported but transcription will be slower. On Windows: Windows 11 or later with a modern multi-core processor, and an NVIDIA GPU for the fastest processing. At least 8GB of RAM and 4GB of free disk space are recommended on both platforms.
How long does it take to transcribe a meeting?
With GPU acceleration (Apple Silicon on Mac, NVIDIA CUDA on Windows), a typical 1-hour meeting transcribes in 1-2 minutes. Without a supported GPU, expect around 5-10 minutes depending on hardware and model. Transcription runs in the background, so you can keep working, and dictation can jump the queue at any time.
Do I need an internet connection?
Only to download the AI models during setup and to activate your license. After that, everything works offline. Record, transcribe, summarize, and dictate on a plane, in a secure facility, or anywhere else.
Can I ask questions about my meetings?
Yes. Open any recording and use Chat with this meeting to ask questions like "What were the action items?" or "What did we decide about the budget?" Answers cite the moments they come from, so you can click a timestamp to jump straight to that point. The chat runs on a local model.
Can I customize the summary format?
Yes. Choose from built-in templates (Action Items, Prospecting, Project Discussion, General) or write your own. You can also edit any summary or transcript line after the fact.
Can I use my license on multiple devices?
Each license covers one computer, Mac or Windows, your choice. Need more? Pick the number of devices when you buy, and manage your activations from your account dashboard at any time.

Your next meeting is a good test. So is your next email.

Tap image to zoom