Turn text into spoken audio, entirely on-device. No cloud, no internet required after initial model download.
Status: Wireframe / MVP prototype
Platforms: Android-first, iOS-aware
Stack target: On-device AI voice synthesis (ONNX runtime)
- Live prototype: Open
prototype.htmlin any browser - PRD
- Onboarding spec
- Data models
- Build plan
- Wireframe notes
- Create account (or skip — local-only option always visible)
- Device check → model recommendation
- Download voice model (~420 MB, one-time)
- Import content (paste text, PDF, article link, EPUB)
- Preview + choose voice with playback
- Generate audio offline
- Save / Share / Regenerate
- Saved library with share, download, delete
| File | Purpose |
|---|---|
prototype.html |
Interactive clickable wireframe — full onboarding + generation flow + saved library |
prd.md |
Product requirements — MVP scope, gates, decisions |
onboarding-spec.md |
Detailed onboarding flow spec |
data-models.md |
User, device, model, generation data structures |
build-plan.md |
Technical build plan — architecture, phases, constraints |
wireframe-notes.md |
Wireframe design notes |
wireframe.svg / wireframe.png |
Visual wireframe |
logo.jpg |
App logo |
- Android-first, iOS-aware
- Balanced (single model tier for MVP — no Lite/Pro)
- First demo path: Account → Device Check → Model Recommendation → Download → Paste Text → Choose Voice → Generate → Play → Export
- Always-visible "Continue local-only" option
- Offline-first: documents and audio never leave the device