◀ BACK vaqlo macOS · MIT

MACOS · FREE · OPEN SOURCE

vaqlo

your meetings, transcribed on your mac.

vaqlo records your microphone and everything playing through the speakers — zoom, meet, a call in the browser — transcribes it on-device with whisper, separates and names the speakers, and writes searchable markdown notes with local summaries. no cloud, no subscription, no accounts. audio never leaves your mac.

~9 MBDOWNLOAD
7LANGUAGES
0CLOUD
0ACCOUNTS
UNIVERSAL DMG · macOS 15+ release notes view source download dmg →

WHAT IT DOES

CAPTURE

microphone + system audio via core audio taps. start by hotkey, menu-bar icon, or control center. auto-detects meetings when an app grabs the mic — with per-app rules so your dictation tools never trigger it.

TRANSCRIBE

bundled whisper.cpp on the apple neural engine / metal, with voice-activity detection. auto language detection. nothing is uploaded.

SPEAKERS

on-device diarization splits voices. name someone once and a local voice library recognizes them in future meetings. titles and participants come from your calendar (incl. a connected google account).

NOTES

day/week timeline, an audio player with per-track mute and follow-along, full-text search, and local LLM summaries — tl;dr, decisions, action items — exported as markdown.

HOW IT WORKS

01
start recording

press ⌥⌘ R, click the menu-bar icon, or flip the control-center toggle. or let vaqlo notice the meeting and offer to record.

02
both sides, two tracks

your voice and the system audio are captured separately, so “you” and “the call” stay distinct. chunked to disk — a crash loses seconds, not the meeting.

03
transcribe & summarize

manually or on a schedule. whisper transcribes, voices get names, a local model writes the summary. audio then moves to the trash; the transcript stays.

SPECS

REQUIRES
macOS 15 Sequoia or later · arm64 or x86_64
DISTRIBUTION
signed + notarized DMG via GitHub Releases
AUTO-UPDATE
Sparkle, EdDSA-signed, 24h check interval
MODELS
Whisper / LLM weights downloaded on first use from Hugging Face (~0.5–2.4 GB), stored locally
LICENSE
MIT — source on github
TELEMETRY
none. no servers, no analytics. audio and transcripts never leave your mac.
LANGUAGES
uk, en, fr, es, pt, de, it
PRICE
free.
vaqlo · part of panic-kit · hello@panic-kit.com · MIT on github
vaqlo loaded PANIC-KIT © 2026