Kuberwastaken / megaphone

What it does

• Fully on-device transcription. Apple's speech model processes your audio directly on your Mac. There is no transcription API, API key, or internet connection required.
• Results as soon as you stop speaking. Megaphone streams audio into the analyzer while you're talking, so most of the work is already done by the time you release the shortcut.
• Hold-to-talk or toggle mode. hold Fn to dictate, or press Command-Fn to start and stop recording. Both shortcuts can be changed.
• Smart on-device cleanup. Apple's Foundation Models framework removes fillers, resolves self-corrections, and fixes punctuation without uploading your transcript. Setup checks whether Apple Intelligence is ready and can open its System Sett
• Inline AI (Alpha). hold your dictation shortcut and begin with “Hey Megaphone” to ask a question, generate text, or transform the sentence you just dictated (for example, “make that formal”). Megaphone uses recent text and active-app conte