Hackathon · 2026 · Voice AI Hack · London 2026
TrueVoice
Clinical voice intelligence platform that flags discrepancies between patient words and vocal biomarkers in real time during medical consultations.
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
Challenge
Patients routinely minimize symptoms during consultations, leaving clinicians with an incomplete picture. Voice biomarkers carry that signal but sit on a separate diagnostic surface from the live transcript.
Approach
Built a real-time clinical platform fusing medical-grade transcription with three parallel voice biomarker streams (distress, mood/energy, affect), reconciled by a concordance engine that surfaces minimisation flags in under a second.
What it does
- Concordance Engine. Matches minimisation phrases against live biomarker evidence to flag word/voice discrepancies in <1s.
- Three Biomarker Streams. Parallel distress, mood/energy, and affect signals via Thymia Sentinel.
- Two Consultation Modes. Telehealth (separate devices) and in-person (single microphone) with diarized transcription.
- Evidence Reports. End-of-consultation summaries synthesised by Claude with cited concordance gaps.
Gallery
Photos
