Talking is faster than typing — but raw dictation is messy, and different writing contexts need completely different tones and structures. FiavaionDictate solves both problems at once. You speak, the AI cleans up the dictation, then formats it for your chosen output mode: Free Form — light cleanup, preserves your voice Formal — professional, polished prose Friendly — warm, conversational tone Email — ready-to-send structure Developer / Claude Code — structured as a precise AI prompt …and more Other things worth knowing: 60+ voice commands — punctuation, editing, formatting, all hands-free Diff view — see exactly what the AI changed Ghost text predictions as you speak "Diagram this" — generates Mermaid diagrams from voice Macro recorder for phrases you repeat constantly AI providers: Claude API, Gemini, or local Ollama (fully offline) No account, no subscription, no data leaves your machine One double-click to start (Python local server, Windows/Mac/Linux) It started as an accessibility project — speech is often the most natural input for people who find typing difficult — but the productivity case turned out to be just as strong. Happy to answer questions from either angle.

FiavaionDictate

Speak. AI refines it. Get the output format you need.

Speak. AI refines it. Get the output format you need.