GitHubDiscord

macOS Installation

Yumi is fully compatible with macOS, operating natively on both Intel (x86_64) and Apple Silicon (M1/M2/M3) architectures.

⚡ One-Line Quick Install

Open a standard Terminal window and run the automated bash installation script:

curl -LsSf https://raw.githubusercontent.com/CodeNeuron58/Yumi/main/install.sh | sh

What this script does:

  1. Ensures Homebrew is installed.
  2. Installs Python 3.12 (if missing).
  3. Installs the uv package manager securely.
  4. Clones the Yumi repository to your home folder.
  5. Sets up the virtual environment with CPU-specific PyTorch wheels.
  6. Pins a global shell alias so typing yumi works everywhere.

🛠️ macOS Audio Permissions & Dependencies

Because of macOS's strict sandbox security, a couple of steps are required to ensure audio capture and keyring storage work flawlessly:

1. Terminal Microphone Permissions

If you run Yumi through a terminal emulator (like Terminal, iTerm2, or VS Code), that app must be granted permission to access your microphone.

  1. Go to System Settings > Privacy & Security > Microphone.
  2. Ensure your terminal app (e.g. Terminal or iTerm2) is toggled ON.
  3. If Yumi still fails to hear you, try running her from a clean terminal window so macOS can trigger the native permission prompt.

2. Apple Silicon (M1/M2/M3) Optimization

Apple Silicon chips include hardware matrix coprocessors (AMX) that accelerate local speech recognition.

  • Performance: If using Local Whisper on Apple Silicon, execution is extremely fast and light.
  • Keyring Security: Yumi securely commits your API keys to the macOS Keychain Access utility. The first time she queries the keychain, macOS will display a secure system dialog asking for your password or TouchID to authorize access. Select Always Allow.

Proceed to the Linux Installation or Core Senses sections to learn more!