Quill (Citedy fork)

Menu-bar meeting recorder for macOS: mic + system audio as two tracks, then transcription and a GPT summary in a native result window.

Based on digimata/quill, extended for Russian meetings, OpenAI Whisper, and a proper post-stop UI.

Features

• Menu bar app — no Dock icon (LSUIElement)
• Dual-track recording — mic.caf (you) + system.caf (them)
• OpenAI Whisper transcription (whisper-1, language configurable, default ru)
• Result window on Stop:
  • spinner while transcribing / summarizing
  • full transcript
  • gpt-4o-mini summary + key points
  • copy all / open in Mail / open session folder
• CLI: quill rec, quill stop, quill show, quill doctor

Requirements

• macOS 15+
• Apple Silicon recommended
• OpenAI API key (OPENAI_API_KEY)

Install (DMG)

1. Open Quill-1.1.0.dmg from Releases
2. Drag Quill.app to Applications
3. Launch Quill (menu bar shows Quill)
4. Put your key in ~/.config/quill/env:

mkdir -p ~/.config/quill
cat > ~/.config/quill/env <<'EOF'
OPENAI_API_KEY=sk-...
EOF
chmod 600 ~/.config/quill/env

5. Optional config ~/.config/quill/config.json:

{
  "recordings_dir": "~/Downloads/Recordings",
  "transcription": {
    "enabled": true,
    "engine": "openai",
    "model": "whisper-1",
    "language": "ru",
    "api_key_env": "OPENAI_API_KEY"
  }
}

Build from source

git clone https://github.com/nttylock/citedy-quill.git
cd citedy-quill
swift build -c release
./scripts/package-app.sh
./scripts/package-dmg.sh

Produces:

• dist/Quill.app
• dist/Quill-1.1.0.dmg

Usage

1. Start Quill (menu bar label Quill)
2. Start recording → grant mic + system audio if prompted
3. Stop recording → result window opens automatically
4. Wait for transcript + summary → copy or email

Sessions: ~/Downloads/Recordings/<yyyy.MM.dd-HHmm>/ (configurable)

File	Contents
mic.caf / system.caf	audio tracks
meta.json	timestamps / offsets
transcript.json / transcript.md	transcript
summary.md	gpt-4o-mini summary
transcribe.log	pipeline log

CLI

quill                 # menu-bar daemon
quill doctor          # permissions + API key
quill rec             # start recording (daemon must be running)
quill stop            # stop + transcribe
quill show            # show control window
quill quit-daemon
quill install --launch-at-login

Privacy note

Unlike the original on-device Parakeet path, this fork uploads audio to OpenAI for Whisper and sends transcript text to gpt-4o-mini for summarization. Keep that in mind for sensitive meetings.

License

Same as upstream digimata/quill unless otherwise noted in the original repo. Citedy fork changes are provided as-is for personal / product use.