A Mac turns meeting audio waveforms into a structured note beside recording controls.

Earnote: Private AI Meeting Notes, Recorded on Your Mac

Earnote records meetings and lectures on Apple silicon, transcribes audio locally, and turns the conversation into structured notes you can send to your existing apps.

Key takeaways
  • Records Mac microphone and system audio, transcribes locally with Apple Speech or WhisperKit, and creates structured notes with a selectable local or cloud AI model.
  • Adds call detection, pause and resume, timestamp-linked playback, speaker separation, transcript search, categories, lecture study tools and exports to notes and task apps.
  • The local model keeps note generation on-device; optional cloud providers receive transcript text. Optional iCloud sync covers notes and transcripts, while audio remains on the Mac according to the project.
  • Requires Apple silicon M1 or newer, macOS 15+, and at least 8 GB RAM; Apple Speech and Apple Intelligence need macOS 26+. The current v0.9.21 release provides an official DMG; Earnote is in beta.

A meeting recorder that writes the follow-up

Earnote is a native Mac app for turning meetings, calls and lectures into notes you can use. It records audio on the Mac, creates a transcript, and asks an AI model to organize the important points, instead of making you replay an hour-long conversation to find one decision.

You choose separately where transcription runs, which model writes the note, and where the result is saved. A local transcript does not automatically mean that the note-writing model is local too.

Capture system sound, not just your microphone

Earnote can record your microphone and the sound coming from the Mac at the same time. It uses macOS Core Audio process taps for system audio, so you do not need a virtual audio device such as BlackHole to capture the other people in a call.

It can recognize calls in apps such as Zoom, Teams, Webex, FaceTime and Slack, then offer to record with a small prompt. You can pause for an off-the-record moment; the project says nothing is saved while paused, and recording can stop when the call ends.

Transcribe and summarize on your Mac

Transcription stays on the Mac, using Apple Speech on macOS 26 or later, or Whisper through WhisperKit. For writing notes, Earnote defaults to an included on-device model: download it once, then use it privately and offline. The first model download needs internet and about 2.3 GB of free space.

You can instead choose Apple Intelligence, Ollama, LM Studio, Claude, OpenAI, Gemini, Mistral, an OpenAI-compatible server, or an existing Claude Code or Codex subscription. If you select a cloud AI service, transcript text is sent to that provider.

Give each kind of conversation its own structure

Create categories such as lecture, interview or client call, each with its own icon, color, summary instructions and export targets. A setup assistant guides permissions, model download, AI provider and destinations, without requiring terminal commands.

A lecture category can ask for key concepts and exam hints, while a client-call template can emphasize decisions and follow-up tasks. Earnote can also turn lecture content into question-and-answer flashcards inside the note.

Keep timestamps, playback and search together

Each note has a slim audio player, and its timestamps jump to the corresponding moment in the recording. You can place the transcript beside the note and follow along during playback to check a name or exact phrase.

Earnote separates speakers as you and others by comparing microphone and system audio. Search works across transcripts, a global shortcut can start or stop recording from another app, and recordings longer than three hours are processed in chunks.

Send notes and tasks to the apps you already use

Notes can be exported to Notion, Obsidian, Logseq, Apple Notes, a Markdown folder, Bear or Craft. Tasks can go to Apple Reminders, Things or Todoist, so action items land in the manager you already check.

Menu-bar controls let you start, pause or stop a recording and see recent notes without keeping the main window open. You can also drag an audio file into Earnote and use the same transcription and note workflow.

Know what stays local and what does not

Audio and transcripts are stored in Earnote's application-support folder. Optional iCloud sync keeps notes and transcripts in your private iCloud database; according to the project, audio itself does not leave the Mac. With the built-in local model, note generation also stays on-device.

If you choose a cloud model, transcript text is sent to that provider and the app shows a notice. API keys are stored in the macOS Keychain. Check the provider before recording confidential material, and get participants' consent before capturing a conversation.

Made for lectures as well as meetings

A semester overview connects lectures from one subject, highlights recurring topics and exam hints, and offers a simpler-explanation switch for technical material. Study notes can be saved as PDF or printed; flashcards can be exported as CSV for Anki.

Long recordings are handled in chunks and long transcripts are summarized in stages. That is useful for a three-hour class, but Earnote is still beta software. Keep original recordings for important sessions and review summaries rather than treating them as a perfect record.

Check the Mac requirements and language limits

Earnote requires macOS 15 or later on Apple silicon, starting with M1, and at least 8 GB of RAM. Budget about 2.3 GB of free space for its local AI model. Apple Speech and Apple Intelligence require macOS 26 or later; WhisperKit is the alternative transcription path described by the project.

The app interface currently follows your Mac in English or German. The language used to write notes is selected separately in Settings, so it does not have to match the interface. The latest official release checked for this article is v0.9.21, and the project labels Earnote as beta.

Install Earnote from the official v0.9.21 Release

We did not find a matching Mac App Store listing for this app and developer. The direct installation path is the official Earnote.dmg download; version details are on the v0.9.21 Release page.

Open the disk image and drag Earnote into Applications. The project says the app is signed with an Apple Developer ID and notarized, so a normal release should open with a double-click. There is no need to build from source when this installer is available.

MIT-licensed source, beta software, and recording consent

Earnote's application source is under the MIT License. External AI providers and models have their own terms. See the official repository and README for current feature notes, setup guidance and privacy details.

Recording laws and workplace policies vary. Tell people when you are recording and obtain consent where required; Earnote itself displays a reminder. For sensitive conversations, check the provider setting and review the resulting note before sharing.

Project facts

License
MIT
Platform
macOS 15 or later; Apple Silicon M1 or newer; 8 GB RAM recommended
Written in
Swift and SwiftUI; local transcription through Apple Speech or WhisperKit
Version reviewed
v0.9.21 (latest official GitHub Release checked 2026-09-24)

Install

# Download the official v0.9.21 DMG:
open https://github.com/louiskl/Earnote/releases/download/v0.9.21/Earnote.dmg

# Open the DMG and move Earnote.app to /Applications.

License summary

Earnote application source is MIT-licensed. Selected AI models and external providers have their own terms and privacy policies.