AIYO Whisper

A native macOS app for local LLM voice transcription. Dictation and speech-to-text that run entirely on-device. No audio leaves the machine, no cloud API, no per-minute billing.

The idea

Most dictation tools ship your voice to a server. AIYO Whisper runs the whole pipeline locally on top of whisper.cpp, so transcription is private by construction. You bring the model; the app handles capture, inference, and clean text output into any field.

What it does

• On-device transcription, voice → text with no network round-trip.
• Native macOS surface: SwiftUI app, system-level dictation feel.
• Local models, whisper.cpp backend, MIT-licensed and open.

Status

Open source (MIT) at Aiyo28/aiyo-wisper. The on-device transcription space is now being commoditized by Apple's own SpeechAnalyzer stack, so AIYO Whisper stands as a clean, private, BYO-model reference implementation rather than a competing engine.