EqualVoice

Inspiration

The inspiration for EqualVoice stems from a simple, yet profound realization: communication is a fundamental human right, yet standard technology still leaves millions behind. We wanted to build an accessible, smart wearable that breaks down the persistent communication barriers faced by the deaf and hard-of-hearing community. By transforming hand gestures into instant speech, we wanted to ensure that anyone utilizing sign language can seamlessly interact with the world around them anytime, anywhere.

What it does

EqualVoice is an AI-powered, edge-computing wearable translation device designed as a smart glove and companion glasses system. The wearable glove tracks complex hand orientations and individual finger flexes in real time. When a user forms a gesture or signs a sentence, the onboard system processes the movement data instantaneously, translating the sign language into clear, spoken audio or digital text. This allows for fluid, bi-directional conversations between signers and non-signers without requiring an intermediary translator.

How we built it

We developed the prototype entirely using low-cost, high-efficiency hardware to keep it accessible and truly wearable. The brain of the project is the ESP32-CAM module, which acts as our main controller, utilizing its built-in Bluetooth capabilities to handle audio routing. We integrated 2.2" flex sensors across the primary active fingers (Thumb, Index, and Middle) to record bending angles, alongside an MPU-6050 IMU sensor to capture 3D motion tracking, tilt, and acceleration data. The hardware is powered wirelessly using a lightweight 1200mAh Lithium-Polymer battery managed by a TP4056 charging module with a built-in battery management system (BMS) for safety. The firmware was written using C/C++ to read raw analog and I2C inputs, map the values, and process them locally for fast performance.

Challenges we ran into

One of our biggest hurdles was the severe limitation of hardware resources. The ESP32-CAM is incredibly capable, but it is very tight on available GPIO pins because the camera module uses most of them. Mapping three separate analog flex sensors and an I2C-based IMU sensor required meticulous pin configuration and optimization to avoid conflicts. Additionally, finding the perfect compromise between a lightweight form factor and reliable portable power took a lot of sourcing and component testing until we could secure a flat, stable Li-Po configuration.

Accomplishments that we're proud of

We are incredibly proud to have built a fully functional, self-contained wearable prototype from scratch within the tight constraints of a hackathon timeline. Achieving zero-latency data readings from multiple hardware sensors simultaneously on a low-power microcontroller was a major technical milestone for us. More importantly, we succeeded in designing a piece of technology that balances hardware functionality with practical everyday comfort, moving a massive step closer toward accessible tech for social good.

What we learned

This project taught us an immense amount about edge computing, hardware hacking, and the nuances of physical computing. We learned how to deeply optimize code for microcontrollers where memory and pins are at a premium. Beyond the engineering side, building EqualVoice gave us a much deeper understanding of user-centric design—reminding us that the best assistive technology must be lightweight, safe, discreet, and intuitive enough to feel like an extension of the human body.

What's next for EqualVoice

The next step for EqualVoice is transitioning from hardware baseline tracking to advanced software classification. We plan to build out a robust gesture recognition library using localized micro-machine learning (TinyML) models running directly on the ESP32 chip. We also aim to scale up the number of flex sensors to accommodate all five fingers for complex spelling libraries and refine the mechanical design into an ultra-slim, consumer-ready smart textile glove.

Built With

• dart
• flutter
• mediapipe
• sign