karansinghgit/speaktype — reverse-engineered prompt

Build me a simple macOS app called SpeakType that lets me dictate text anywhere on my Mac without sending audio to the cloud.

I want it to work like this, I press a global hotkey, fn by default, speak for a few seconds, release the key, and the transcribed text gets pasted into whatever text field I was using. It should run fully offline using a local Whisper model, with a settings screen where I can download or choose the AI model, change the hotkey, and see what permissions are needed.

Please make the app feel native to macOS, with a small clean SwiftUI interface, microphone and accessibility permission handling, clear status messages like listening, transcribing, model loading, and ready. It should be optimized for Apple Silicon and macOS Ventura or newer.

Also include basic tests and simple developer commands to build, run, test, and package a DMG. Look up current docs online if you need to.