Hackathon · 2026 · Voice AI Hack · London 2026
TrueVoice
A platform that listens during medical consultations and flags when a patient sounds worse than their words say.
Overall Winner - Voice AI Hack London 2026
First place across the full event in the Voice & Medical track sponsored by Thymia and Speechmatics.
TypeScript · Next.js · Python · Claude API
The problem
Patients often play down their symptoms when talking to a doctor. The signs are there in their voice, but the doctor has no way to catch them during the conversation.
What I built
I built a platform that transcribes the consultation while reading three signals from the patient's voice: distress, mood, and emotion. When the words and the voice do not match, it flags it to the clinician in under a second.
What it does
- Catches the Mismatch. It compares what the patient says with how their voice sounds, and flags the gap to the clinician in under a second.
- Three Voice Signals. Distress, mood, and emotion are read from the voice in parallel, using Thymia's Sentinel models.
- Works in Any Consultation. Telehealth with separate devices or in person with one microphone, and the transcript keeps track of who is speaking.
- Reports With Evidence. At the end, Claude writes a summary that points to every moment where the words and the voice did not match.
Photos
Gallery
The full engineering is on GitHub.