Sonora is a Windows desktop speech studio that runs fully offline after models download, no cloud API for daily use. • Multi-engine local TTS with a studio UI and custom voices • Speech-to-text with optional speaker labels • Global dictation: hotkey voice typing into any focused Windows field • Localhost HTTP API so scripts and AI agents can call TTS/STT on your machine GPU-friendly, works on CPU too. Privacy-first voice for makers who don’t want every utterance shipped to a SaaS.