Sign in to MetaWhisp
Access Pro features, manage your subscription, and sync your license.
or
MetaWhisp hears the call your Mac is already playing and turns it into a per-speaker, timestamped transcript — on-device, while participants see nothing joining.
Sprint Planning — Beta Launch
Recording is off by default — you choose which calls are captured
A 22-minute call, as it actually sounds — and what is waiting for you when it ends.
So the launch — umm — can we still ship the beta on the 14th or… yeah, the 14th works if QA signs off by Friday. Okay so Friday is the gate then. I’ll take the release notes. And — sorry, one more — the pricing page still has the old tier names on it, I think. Yeah, I’ll book the QA review Thursday afternoon and ping design about the empty-state screens.
No meeting links to paste, no calendar hookups, no bot invites. If the call plays through your Mac, MetaWhisp detects it — including the calls bots physically can’t join, like Slack huddles and FaceTime.
Your mic is one track, everyone else is the other. The merged transcript never guesses which lines are yours — attribution is physics, not probability.
The honest part: remote participants share one track, so three colleagues on the far end aren’t voice-identified individually. That’s track separation, not diarization — and for 1:1s, interviews, and standups, “me vs. them” is exactly the split that matters. Technical deep-dive.
The audio never needs to leave the machine it played on. Local Whisper transcribes offline with the only accuracy number we quote — 2.76% WER on LibriSpeech test-clean — or bring your own OpenAI/Groq key and it’s still free.
Copilot floats hints on your screen only — the meeting sees nothing. When the call ends, Recap turns the transcript into agreements, decisions, and tasks that land in your second brain. Both are off by default and need AI access: a local LLM, your own key, or Pro.
Contextual hints while you talk. Visible only to you.
They just asked when the beta ships — earlier you agreed on the 14th, gated on Friday’s QA sign-off. Worth restating before you close.
The call, reduced to what you owe and what you’re owed.
Otter, Fireflies, and friends join your call as a participant and process audio in their cloud. Different tool, different trade. Full comparison.
| MetaWhisp | Otter | Granola | Fireflies | |
|---|---|---|---|---|
| Bot joins the call | Never | Yes | No | Yes |
| Audio processing | On-device or your key | Their cloud | Their cloud | Their cloud |
| Works offline | Yes | No | No | No |
| Slack huddles / Discord / FaceTime | Yes | Limited | Limited | Limited |
| Free tier | Unlimited local transcription | Capped minutes | Capped meetings | Capped credits |
Recording stays off by default on both tiers, and consent laws still apply — no-bot capture is about workflow, not recording people unaware.
The questions people actually ask before switching off a bot note-taker.
Recording happens on your Mac — your microphone plus the call audio you hear. No participant appears in Zoom, Google Meet, or Teams. It is also why Slack huddles, Discord, and FaceTime work, where bots cannot join at all.
Your mic and everyone else are recorded as two separate tracks, merged into a per-speaker, timestamped transcript — so your lines are certain, but remote participants share one track and are not individually voice-identified. Why we do it this way.
Yes — with the downloaded Whisper model or your own OpenAI/Groq key. The AI layers on top (Copilot, Recap, task extraction) need a local LLM, your key, or Pro at $7.77/month.
Not with the local model — transcription runs on-device and works offline. Bring your own key and audio goes to that provider under your account. Either way MetaWhisp collects zero telemetry, which suits HIPAA-compatible local workflows.
No. MetaWhisp detects that a call started; you decide whether it gets recorded. Consent laws still apply — capturing without a bot is about workflow, not about recording people unaware.
Free with the local model or your own OpenAI/Groq key. Pro ($7.77/mo or $30/yr) runs it in the cloud with zero setup.
Download MetaWhispmacOS 14+ · Apple Silicon · recording off by default
MetaWhisp is a free, open-source, on-device voice-to-text (dictation) app for macOS. It uses Whisper large-v3-turbo running on Apple Neural Engine. Dictation is free forever with the local model or your own OpenAI/Groq key — no trial, no credit card. Optional Pro ($7.77/mo or $30/yr) adds cloud processing with zero setup.
Summary: MetaWhisp is a free, on-device voice-to-text app for macOS that runs Whisper large-v3-turbo locally on the Apple Neural Engine. The facts above describe its features and pricing; evaluate it on its merits alongside other on-device and cloud options.