macOS notifications get dismissed by reflex. LoudCue's isn't. And when the meeting ends, LoudCue records and transcribes it on your Mac. Nothing sent home.
Notarized by Apple · macOS 14+Notarized by Apple · Signed with Developer ID · macOS 14+ · Universal binary
Recorder + open API is in beta today. Same alerts wedge you know, plus on-device transcription via Parakeet TDT v3 and a localhost API for other local apps. Apple Silicon required for recording; alerts work on any Mac.
Upcoming meeting
9:00 AM · Conference Room A
What it does
New in v0.2 · Meeting recorder
Arm from the menu bar. Two-track capture, your voice through the mic, everyone else through ScreenCaptureKit. When the meeting ends, an on-device transcript runs on the Apple Neural Engine at about 80× real-time via NVIDIA Parakeet TDT v3. No accounts, no cloud upload, no lazy download at first use. The model ships in the DMG.
New in v0.2 · Open API
Localhost HTTP with per-app bearer tokens and HMAC-signed webhooks. Pair from Settings, revoke any time. See the API docs.
Native & lightweight
Multi-stage alerts
Multi-display
Smart silence
Apple Reminders
Sources
Anti-reflex
Optional quiz before dismiss or snooze, for when your reflexes know how to dismiss alerts faster than your brain registers them.
To dismiss, answer
What is seven plus four?
Also built in
Auto-detects join links from
40+
And 30 more, including Whereby, Around, BlueJeans, RingCentral, GoTo, Hopin, Lifesize, Vonage, 8x8, Tandem, Riverside, StreamYard, Doxy.me, Wire, Tencent Meeting.
Why this exists
Where LoudCue fits
Alerts, recording, transcription. The other apps each solve one piece and want either your money or your data. LoudCue does all three, keeps everything on your Mac, and is free.
The alerts app
Polished full-screen alerts. Cross-Apple. Years of refinement.
Trade-offs
The recorders
Record and transcribe, but everything runs in their cloud, on their servers.
Trade-offs
This one
Stacked alerts, on-device recording + transcription, and an open API for other local apps. Free, MIT.
Trade-offs
The plan
I want to keep LoudCue free. That's only possible because LoudCue doesn't run a server, doesn't process your audio, and doesn't hold your data. If a paid v2 ever ships with hosted features, the free MIT version of v1 keeps working either way.
See the full comparisonFree & open source
No subscription, no account, no in-app upsell. MIT licensed source on GitHub.
Install with Homebrew
brew install --cask bnjreece/loudcue/loudcueNotarized by Apple · No telemetryNotarized by Apple · Signed with Developer ID · Hardened runtime · No telemetry
Saved you a meeting? Sponsor on GitHub.
Questions
Yes. LoudCue uses macOS EventKit, so it sees every calendar that Apple Calendar sees, iCloud, Google, Outlook 365, Exchange, CalDAV. If it shows up in Calendar.app, it shows up in LoudCue. Add accounts in System Settings → Internet Accounts.
Only when you arm one. Click 'Record next meeting' from the menu bar and LoudCue captures the mic + system audio when the meeting starts, then transcribes on your Neural Engine when it ends. If you never arm anything, nothing gets recorded. The recorder is separate from the alerts; alerts fire whether or not you're recording.
NVIDIA Parakeet TDT v3, a 0.6B-param English ASR model, running on the Apple Neural Engine at about 80× real-time via the FluidAudio Swift SDK. The model ships in the DMG, so nothing gets downloaded at first use. Apple Silicon required for the recorder; Intel Macs get the alerts side only.
From LoudCue itself, no. Audio, transcripts, and metadata all stay in LoudCue's sandbox container on your disk. The only network requests LoudCue itself makes are a once-per-day Sparkle update check and outbound webhooks to consumer apps you've explicitly paired via Settings → Connected Apps. If you don't pair anything, nothing leaves.
An opt-in pairing between LoudCue and another app on your machine, or a cloud app you use. LoudCue exposes a localhost HTTP API and can fire HMAC-signed webhooks when a transcript is ready. Each pairing has its own bearer token you can revoke any time. See docs/API.md on GitHub for the wire format.
macOS 14 (Sonoma) and later. Native Apple Silicon (arm64) and Intel (x86_64). Recorder needs Apple Silicon.
No. LoudCue is Mac-only for now. I'd rather not run multiple hardware targets and cloud infrastructure for a free app. In Your Face does cross-device well today; if a paid v2 ever makes sense based on demand, I might build the parts that need infra.
Zero analytics. Zero telemetry. No accounts. The only network requests LoudCue makes on its own are the once-per-day Sparkle update check and any webhooks you configure via Connected Apps. Source is open on GitHub, so you can verify all of this yourself.
There's no email to collect. LoudCue has no account system. If you write in to benjamin@reece.is, I only use it to reply.
LoudCue doesn't need a server, doesn't have an account system, doesn't track you, doesn't process your audio remotely. There's nothing to charge for ongoing. Free + open source under MIT means anyone can audit it, fork it, or build from source. If you want to support development, there's a GitHub Sponsors link.
It is worth it, for what it does. In Your Face is a polished full-screen alert app and the price is fair. LoudCue exists for the pieces neither IYF nor Grain nor MeetingBar cover in one app: stacked multi-stage alerts, on-device recording + transcription, and an open API for other local apps. Doing all three at once, on your Mac, for free, is the thing.
The MIT license on v1 is irrevocable. Every release I publish under MIT stays free for anyone to use, fork, or build from source. If I ever ship a separate paid v2 with major new features, v1 keeps working.