MimicScribe
Show HN: On-device transcriber that's 97% accurate at identifying speakers
Last verified:
What is MimicScribe?
MimicScribe is a macOS menu bar app that serves as an in-meeting technical assistant and transcription tool. It captures system audio at the OS level, enabling speech-to-text transcription for Zoom, Teams, Google Meet, Slack huddles, and any app that plays sound without joining calls as a bot. The app runs speech-to-text entirely on-device using NVIDIA Parakeet via CoreML, with speaker identification accuracy of 96-98%. It is specifically tuned for requirements gathering, goal tracking, and surfacing prep notes during fast, packed meetings.
Key features include named speaker identification (not
MimicScribe pricing
Pricing model: Freemium
Free tier: $0, no card required - includes unlimited meeting summaries with speakers separated and named, 10 in-meeting briefings per day, 10 meeting follow-up questions & corrections per day, 200 voice typing with grammar correction per day, 10 file transcriptions per day. Requires macOS 15+ and Apple Silicon. Light plan: $48/year ($4/month) saving 33% - includes unlimited meeting summaries, 100 in-meeting briefings per month, 100 follow-up questions & corrections per month, unlimited voice typing, 100 file transcriptions per month, brings your own Gemini key included. Unlimited plan: $120/year ($10/month) saving 33% - includes unlimited meeting summaries, unlimited in-meeting briefings, unlimited follow-up questions & corrections, unlimited voice typing, unlimited file transcription, brings your own Gemini key included. All plans include 30-day money-back guarantee. checkout and billing handled by Stripe.
MimicScribe pros
- 97% accurate transcription with named speakers
- No meeting bots required - captures OS-level audio
- Works with Zoom, Teams, Google Meet, Slack huddles
- 96-98% speaker identification accuracy
- On-device speech processing - audio never leaves Mac
- Keyboard-first design reachable from any app
- Unlimited meetings and summaries on free tier
- No account or email required
- Voice profiles saved across meetings automatically
- 60-minute meeting finishes transcription in about 1 minute
- Action items auto-assigned to right owner
- Time zone conversion for deadlines automatically
- One-click push to Apple Reminders or Calendar
- Semantic search across all meetings by meaning
- Completely offline mode available
- GDPR and CCPA compliant with zero retention
- Bring your own API key option
- 25 languages supported including code-switching
- Tuned against 96-scenario benchmark to eliminate hallucinations
- Minimal memory usage at about 330 MB during meetings
MimicScribe cons
- Only works on macOS 15+ with Apple Silicon
- AI features like summaries require internet connection
- Free tier limited to 10 briefings and 10 follow-ups daily
- Voice typing on free tier limited to 200 words daily
- File transcription on free tier limited to 10 daily
- Accessibility permission needed for voice editing feature
- Does not announce itself to meeting participants
- Speaker attribution requires cloud AI not fully offline
- Subscription billing handled through third-party Polar
- No Windows or Linux version available
Frequently asked questions about MimicScribe
Is it really free?
Yes - free within fair use. Recording, transcription, and summaries with named speakers are unlimited with no account or card required. The live in-meeting assistant including briefings and follow-up questions has a free daily allowance that a subscription lifts. The limits exist to stop abuse, not to meter normal work, including a lifetime ceiling that is generous but not infinite.
Does it work with Zoom, Teams, and other apps?
Yes. Most AI notetakers join the call as a participant bot in the attendee list with a recording badge. MimicScribe captures system audio at the OS level instead, so it works with Zoom, Teams, Google Meet, Slack huddles, or any app that plays sound. Speakers are still identified on-device without anything joining the call. No plugins or browser extensions required.
Does it handle other languages and technical terms?
Yes. The speech model recognizes 25 languages including English, Spanish, French, German, Italian, Portuguese, and more. Meetings primarily in English with occasional code-switching such as a customer quote in Spanish, a German product name, or a French aside transcribe cleanly and stay searchable. Technical vocabulary, proper nouns, and jargon work without a custom dictionary.
What happens if I lose my internet connection?
Transcription always works offline since it runs entirely on your Mac. AI features like text refinement, meeting summaries, and the meeting assistant require an internet connection. If you lose connectivity mid-meeting, transcription continues uninterrupted and AI enrichments are queued until you are back online.
What permissions does MimicScribe need?
Microphone is required for all voice features. System Audio Recording captures audio from video calls for meeting recording - audio only with no screen capture or video. Accessibility is optional and only needed for voice editing. Skip it and meeting recording still works.
Do I need to tell people I'm recording?
Recording laws vary by state and country - some US states and most of Europe require all parties consent. MimicScribe does not announce itself to meeting participants; that is your call. Best practice is to tell attendees at the top of the call and stop recording if anyone objects. If you are in a regulated industry, check with your legal or compliance team before rolling it out.
Will MimicScribe slow down my Mac?
No. MimicScribe holds at about 330 MB during a meeting which is less than a single browser tab, and the heap does not grow as the recording runs long. ML models run on the Neural Engine, not main RAM, so your other apps are not competing for memory.
What runs on-device vs in the cloud?
Speech recognition, speaker separation, and echo cancellation run on your Mac. Speaker attribution, action items, summaries, and real-time suggestions require larger models than what can run locally with acceptable accuracy so these are handled by cloud AI using transcript text only. Your data is never stored or used for training.
How do I get the best results?
The AI uses whatever context you give it. Spend two minutes filling out Your Context including your role and what you care about, and add a reference document such as product positioning, pricing notes, or objection handlers. This turns generic summaries into briefings that actually understand your work. A setup guide with examples is available.
How many devices can I activate我的 license on?
Each license key can be activated on up to two Macs. If you try to activate on a third device, activation will fail with a message indicating the limit has been reached. To free up a slot, deactivate the key on a device you no longer use through Settings or manage activations through the Polar customer portal.