AeroType - Ultimate Local Global Voice Typing Engine

🔒 Absolute Privacy (100% Local)

Zero Network Dependency: All audio data and recognition computing run offline on your physical device.
No Cloud Backdoors: Eliminates audio uploads completely to protect trade secrets and personal privacy.
On-Demand Model Scheduling: Engine and models are fully decoupled; only download the required open-source quantized model upon initial setup.

⚡ Blazing Fast (< 300ms)

Rust Architecture: Built purely in Rust natively, bypassing Python runtimes and eradicating Garbage Collection (GC) stutters.
Hardware Acceleration: Deeply optimized for macOS ANE (Apple Neural Engine) and Windows DirectML.
Streaming VAD Slicing: Smart silence detection enables real-time recognition while speaking; text applies instantly on release.

🪶 Ultra-light

Micro Installer: Exceptionally tiny base executable file size.
Minimal Resource Footprint: Standby physical memory under 50MB, with negligible CPU wake-ups and battery drain.
Invisible Interaction: Runs silently in the menu bar/system tray, no Dock icon clutter, triggered via global shortcuts.

Use Cases

🤖 Claude Code Rapid Long Text Input

When using AI coding assistants like Claude Code, replace keystrokes with voice dictation for complex system architecture requirements or extensive refactoring logic.
Tests show an 80% reduction in output time, making AI interaction actually match your thinking speed.

📞 Frictionless Conference Call Logging

Take notes during business communications and tech meetings without frantically typing.
Enable global long-voice input to accurately transcribe crucial meeting content on the fly, with effortless post-meeting organization.

💬 Instant Replies to Colleagues

Handle high-frequency messages on WeChat, Slack, or Teams without moving your hands from your current workspace.
Press the shortcut to dictate your reply, and watch it instantly send, radically reducing the physical friction of daily communication.

Q&A

Q: Why can't I use it immediately after first opening?

A: On first launch, the app automatically downloads the speech recognition model to your local device. Please wait patiently for the progress bar to complete — the duration depends on your network speed. This is a one-time download.

Q: Can the app be used offline?

A: Yes. Once the model is downloaded, all speech recognition and processing happen entirely on your device — no internet connection is required during use.

Q: How do I start dictation?

A: Press and hold the global shortcut (default: Ctrl+Shift+Space) to start recording, release to stop and insert transcribed text. You can customize the shortcut in the app settings.

Q: Why is the dictation not accurate?

A: Please ensure you select the correct language in the app that matches what you are dictating. Currently only English and Chinese (including Mandarin and Cantonese) are supported — more languages are coming soon.

Q: Does the app work on Windows?

A: Currently macOS is fully supported. Windows support is under development and will be available soon.

The speed of thought is the speed of typing