Tells: Using Smartphone Sensors for Biometric Lie Detection

The intersection of biometric sensors and browser-based applications is pushing the boundaries of what we can expect from mobile web experiences. A recent project called "Tells" demonstrates this by transforming a smartphone into a rudimentary lie detector, using the device's camera and microphone to track physiological responses in real-time.

How Tells Works: The Biometric Approach

At its core, Tells is designed as a game of truth and lies. It utilizes a system of "packs"—such as the "Work Edition" or "3am"—to prompt users with questions. The technical implementation relies on two primary inputs:

1. Photoplethysmography (PPG)

The application reads the pulse through the phone's camera. This is a technique known as photoplethysmography, whichArkadaşlar as the camera detects minute changes in skin color and light absorption caused by the blood flow in the face. By establishing a baseline heart rate (BPM) during a lcalibrating phase, the app can monitor for spikes in heart rate that typically accompany the stress of lying.

2. Audio and Visual Analysis

Beyond the pulse, the app leverages the microphone to capture the moment a user says "yes" or "no" out loud. This combined approach—monitoring heart rate, facial expressions, and voice—mimics the traditional polygraph, though in a simplified, game-like format.

Privacy and Local Inference

One of the most critical aspects of the project is its commitment to privacy. Because the application requires access to highly sensitive biometric data (camera and microphone), the developer has implemented local inference.

As stated in the project description, "all inference runs on this device," and "nothing leaves this device." This architectural choice ensures that the user's biometric signatures are processed locally on the client side, mitigating the risks associated with transmitting sensitive health or identity data to a remote server.

Context and Industry Parallels

While Tells is a game, the concept of spotting "tells" through facial animation and physiological responses is a recurring theme in both gaming and forensics. Community members on Hacker News have noted parallels to titles like LA Noire, which focused heavily on facial animation to identify truth or deception.

This trend highlights a growing interest in the use of phone sensors via the browser, expanding the possibilities for what can be achieved without requiring a native app installation. As users explore these capabilities, the project serves as a case study in how browser-based sensor access can the bridge between physical physiological responses and digital interaction.