SayItDev

On-device AI and voice for Mac. The dev CLI runs Apple Intelligence locally — speak, listen, transcribe, chat, and serve an OpenAI-compatible API. No cloud, no API keys.

Requirements

• macOS 26+ on Apple Silicon
• Apple Intelligence enabled (for LLM / dev "prompt" and --serve only)

Install

brew tap innovatorved/tap
brew install innovatorved/tap/dev
dev --version

Build from source: clone this repo, run make build && sudo make install. See docs/brew-install.md for troubleshooting.

For --listen, grant Microphone and Speech Recognition to Terminal.app in System Settings → Privacy. For --transcribe and --speak, Speech Recognition / output access only (no mic).

Usage

Command	What it does
dev --speak "hello"	Text to speech
dev --listen	Mic → transcript
dev --transcribe file.wav	Audio file → transcript
dev "your prompt"	On-device LLM
dev --serve	Local HTTP server (port 11434)

dev --speak "Hello from SayItDev"
dev --listen
dev --transcribe recording.wav
dev "What is 2+2?"
dev --serve

--transcribe runs fully on-device. Optional flags: --locale en-US, --timestamps (prefix each segment with its time range).

Server endpoints include /v1/chat/completions, /v1/audio/speech, and /v1/audio/transcriptions. See docs/server-security.md before exposing the server on your network.

Docs

• CLI reference
• Homebrew install
• Server security

Credits

Fork of apfel with voice extensions by SayItDev / innovatorved.

MIT — see LICENSE.