Voice Turing Test
Full-stack development lead · May 2025 – Aug. 2025
Voice Turing Test is a public WeChat Mini Program that asks players to listen to multi-speaker conversations and decide whether a responder is human or AI. It turns a research evaluation into a compact game that people can understand in seconds.
From research question to a playable product
Players hear a conversation, inspect the dialogue context, and judge whether the target voice belongs to a human or an AI system. Levels, achievements, combo rewards, referrals, leaderboards, and Chinese-English localization make repeated evaluation feel like a challenge rather than a survey. The experience also builds awareness of increasingly realistic synthetic speech.
Research platform
The product served as the data-collection platform for Human or Machine? A Preliminary Turing Test for Speech-to-Speech Interaction, published at ICLR 2026. I was not a paper author; my role was to lead the platform's product design and full-stack implementation.
My contribution
I independently owned the workflow from Mini Program registration and compliance through UI/UX design, front-end and back-end development, testing, deployment, and operations. The native WeChat front end controls the listening and judgment flow, while Tencent Cloud functions, database, and storage support user accounts, audio resources, scoring, and research-data collection.