I built an open-source voice input for macOS that never uploads your voice

Description and use

Cadenza provides a privacy‑first, cost‑free voice input solution for macOS, with full local processing and optional cloud integration.

Problem

Most voice input tools upload recordings to servers, raising privacy and cost concerns.

Outcome

Cadenza provides a privacy‑first, cost‑free voice input solution for macOS, with full local processing and optional cloud integration.

Workflow

• 1. Hold a configurable hotkey to start recording; a bar shows recording status.
• 2. On‑device inference via sherpa‑onnx using one of five models (e.g., SenseVoice).
• 3. Release the key; the transcription is typed at the cursor, with optional AI polish using a local Ollama instance.