Piper Read Aloud is a local text-to-speech extension. It does not collect,
store, or transmit any personal data, and it contains no analytics or tracking.

• When you choose to download a voice, the extension downloads that voice's
  AI model files (tens of MB) from huggingface.co and caches them locally
  in the browser's Origin Private File System. The available-voices list is
  also fetched from there (with a bundled offline fallback).
• As a fallback for missing bundled files only, speech-engine runtime files
  (ONNX Runtime WebAssembly, Piper phonemizer) may be fetched from
  cdn.jsdelivr.net or cdnjs.cloudflare.com. In normal operation these
  files are bundled inside the extension and are never requested.

Nothing you read, select, or type is ever transmitted anywhere. Speech
synthesis runs entirely on your device.