Azex Speech

Azex Speech is a native macOS voice‑input application aimed at developers and professionals who work in AI or crypto and frequently switch between Chinese and English terminology. It captures spoken input, runs a bilingual ASR engine locally, and then applies three additional correction layers—domain‑specific vocabularies, context‑aware hot‑word extraction from the active window, and personalized pronunciation learning—to produce accurate text that is automatically pasted into the target application.

The software is open‑source and offered under a free tier, running entirely on Apple Silicon Macs with macOS 14 or newer. All audio processing stays on the device, storing only JSON files locally, so no registration or network connection is required. Users can calibrate the system with flashcard‑style readings, and the app continuously adapts to individual accent patterns and frequently used technical terms.

Key features include a built‑in crypto/AI term library of over 1,800 entries, a dashboard that tracks word count and time saved, and a pronunciation‑training module that generates personalized practice cards from usage history. The interface is controlled via a single keyboard shortcut, and the app provides real‑time feedback on recognition and correction results.