SYS/04 · APPLIED ML
VOXMAX
Voice-scoring exploration running on a custom DSP/ML backend. Private, in development, not yet public.
- FRONTEND
- Expo SDK 54
- SCREENS
- 11
- BACKEND
- FastAPI
- AUDIO
- deleted post-analysis
VOXMAX is a voice-scoring exploration, private and still in development: record ~12 seconds of speech, get a VoxScore with four sub-dials and a one-line verdict, delivered on a custom acoustic-feature backend, not a wrapper around an API.
KEY FACTS
- React Native / Expo frontend (SDK 54) across 11 screens.
- FastAPI backend: deterministic acoustic feature scoring plus LLM coaching.
- ~12 s of speech yields a VoxScore (0–100) with four sub-dials and one improvement drill.
- Privacy by design: the backend deletes uploaded audio immediately after analysis.
The React Native frontend
An Expo / React Native app (SDK 54) spanning 11 screens: onboarding, consent, record, analyzing, result, drill, progress, settings, and guard states. The shareable ScoreCard is rendered to an image on-device. Strictly monochrome, lab-instrument aesthetic.
The FastAPI backend
A FastAPI service runs a deterministic acoustic feature-extraction pipeline to compute the score, with an LLM layer producing the coaching verdict and drill. Privacy is enforced server-side: uploaded audio is deleted immediately after analysis, gated behind an explicit consent screen.