Accessibility
Accessibility is not a feature of Aloud. It is the point of Aloud. Every core flow is built to work with the screen off, operated by sound and voice.
What we target
WCAG 2.1 Level AA. Every page is checked against the WCAG 2.1 A and AA rule sets on every push and pull request, and the build fails on a serious or critical violation. Colors meet AA contrast, focus is always visible, motion respects the reduced-motion setting, and controls are keyboard operable and labelled.
How it works without sight
- Status and results are announced through ARIA live regions and spoken aloud.
- The selfie is framed by sound alone: tones speed up as the face centers, with a lighting gate and auto-capture.
- The barcode is found by intensifying beeps, with manual entry as an equal path.
- Voice is a full conversation, with a push-to-talk fallback for networks that block it.
- Every spoken reply is mirrored as on-screen text.
Honest by design
A person who cannot see the screen cannot catch a confident wrong answer, so Aloud states uncertainty in plain words, refuses low confidence readings, and, on deeper skin tones or in low light where analysis is known to be less reliable, lowers its confidence and says so. It describes appearance only, never diagnoses, and makes no medical or treatment claim.
Your data
Nothing is stored on a server. Photos are processed for the session only and are not saved. Aloud does not infer or output race, ethnicity, or health conditions, and does not tie any face or skin data to a name or email.
Report a barrier
If any part of Aloud does not work for you, that is a bug we want to fix. Open an issue at github.com/StephenSook/aloud/issues.