Voice-based Age Verification
PDPC–IMDA Innovation Challenge · ParallelChain Lab · 2024
Step 1Read the phrase aloud before the 20-second timer ends.
Seven sailboats drifted past the quiet harbour as the tide turned.
20sTap the mic to start recording, quick, time is ticking.
A lighter-weight alternative to ID checks.
Many age-verification flows collect more information than they need. ID uploads and facial recognition can answer a simple question, but they also expose sensitive personal data.
For the Singapore Government PDPC–IMDA Innovation Challenge, our two-person team explored a different approach: a user reads a short phrase aloud, a model estimates an age range, and the system returns only that range.
I designed the states around the model.
The model was probabilistic and slightly delayed. Users could not see whether the system had heard them, whether it was still processing, or what to do when it could not return a confident result.
I led the research and interaction design, defining the full state model from recording to retry. The work was less about a single final screen and more about making each transition clear.
The interface had to answer three basic questions.
Is the system listening?
I used a live waveform, countdown, and changing instructions during recording. The user always has a visible signal that the microphone is active and the flow is moving forward.
What happens while it processes?
A silent wait reads as failure. I added a clear processing state after submission so the interface sets expectations while the model runs.
How do I try again?
An unclear estimate should not feel final. Retry states explain what happened in plain language and keep the next action obvious.
The prototype covers the full flow, not just the happy path.
The Figma prototype includes idle, recording, processing, successful result, and retry states. Those transitions are where users decide whether an unfamiliar AI interaction feels responsive and understandable.
Winner of the PDPC–IMDA Innovation Challenge.
The concept won the Singapore Government PDPC–IMDA Innovation Challenge. It demonstrated how an age-verification experience could expose less personal data while still giving users enough feedback to understand what was happening.
This was my first AI interaction project.
It taught me to design the moments around the model, not just the result. State, timing, and recovery are part of the product experience whenever a system is probabilistic or delayed.