Summary:
Add an option to have Lumo's responses read aloud using the text-to-speech engine that's already built into Android and iOS. No server-side synthesis required.

The problem:
Lumo offers voice input on mobile, but there's no way to listen to a response. Users have to stop and read every reply, which makes it awkward to use Lumo hands-free and excludes users who prefer or need audio in support of their individual impairments such as visual strain, dyslexia, and similar situations.

Proposed solution:
Rather than building a separate TTS pipeline, query the onboard text-to-speech engine that every Android and iPhone already ships with. Everything is processed locally on the device, so nothing additional ever leaves the phone or Proton's ecosystem.

Suggested details:
* A play button on responses (or per-message) that hands the text to the system TTS engine
* Playback speed and voice selection, exposed natively by both OS APIs at no extra cost
* Optionally, support for third-party TTS engines the user has installed, so users can pick higher-quality voices or other languages

Why this fits Lumo:
* Zero privacy trade-off: the text is already decrypted on-device to display it; reading it aloud changes nothing about the zero-access encryption model
* No infrastructure cost: it's purely client-side, so it scales without adding server load or cost
* Fits Proton's privacy-first positioning: vocal interaction without shipping user audio or text to a third-party speech cloud

A small feature, but it meaningfully improves accessibility and everyday usability for anyone who prefers listening over reading.