Agentic AI is only as fast as you can prompt it.

Whisper-based, privacy-preserving voice dictation for macOS that boosts your prompts/sec by turning spoken intent into fast, local, on-device text.

🚀 Quick Start     🧠 Features     🛡️ Privacy     📜 License

Features

• Global hotkey — trigger dictation from any app
• Fully local — on-device transcription with whisper.cpp
• Privacy-preserving — no audio, text, or telemetry leaves your Mac
• Built for prompt throughput — turn spoken intent into fast, usable text
• Prompt-aware dictation — adapts to your vocabulary and speaking patterns
• Clipboard-safe insertion — inserts into your focused app and restores your clipboard
• Native menu bar app — lightweight and always ready

Quick Start

make setup-whisper
make run

If you prefer manual setup:

git clone https://github.com/ggerganov/whisper.cpp.git /tmp/whisper.cpp
cd /tmp/whisper.cpp
cmake -B build && cmake --build build --config Release -j
sudo cp build/bin/whisper-cli /usr/local/bin/whisper-cpp
make run

On first launch, grant:

• Microphone — required to record audio
• Accessibility — required to insert text into other apps

Privacy

• All transcription runs locally on-device
• No audio, text, or telemetry leaves your Mac
• Temporary audio files are deleted after transcription
• No analytics, crash reporting, or remote logging

License

MIT

Credits

Built by Zhaoze Wang, with Claude Code and OpenAI.

Powered by whisper.cpp.