Speak Out – An Intelligent Glove for the Vocally Impaired

Developed at Vidyalankar Institute of Technology, Mumbai

As part of our undergraduate engineering program, this project was developed to address a real-world communication gap faced by vocally impaired individuals. The goal was to design a wearable device that could convert hand gestures into understandable text or speech — bridging the gap between sign language and everyday communication.

What Is "Speak Out"?

Speak Out is a smart glove that recognizes hand gestures made in sign language and translates them into readable or audible output via a smartphone. It combines sensor technology, machine learning, and cloud computing to act as a live translator for those who cannot speak but use their hands to communicate.

How It Works

• Input: The glove uses a combination of sensors:
  • Flex sensors to detect finger bends
  • Contact sensors to sense palm contact
  • MPU6050 gyroscope for wrist motion tracking (e.g., letters like J and Z)
• Transmission: An ESP8266 Wi-Fi module sends data from the glove to a cloud database in real-time.
• Processing: A machine learning model (Random Forest algorithm) processes the gesture data and predicts the intended word or letter.
• Output: A mobile application fetches the prediction from the cloud and displays the text or converts it into speech.

My Role & Contributions

• Interfaced and calibrated sensors (flex, contact, and gyroscope) with the microcontroller
• Designed circuit logic and voltage-divider networks for resistance-based sensor inputs
• Integrated ESP8266 Wi-Fi module for real-time data transmission
• Contributed to building and training the ML model for gesture recognition
• Developed the data flow pipeline from glove → cloud → app
• Participated in multiple testing cycles for gesture accuracy and sensor consistency

Why It Matters

Sign language is a rich and expressive form of communication — but most people don’t understand it. With Speak Out, we aimed to create an intuitive, low-cost wearable that helps vocally impaired individuals express themselves clearly in public spaces, hospitals, schools, and more.

“Rather than asking the world to learn sign language, we built a tool that speaks it for you.”

Tech Stack

• Hardware: Flex Sensors (x5), Contact Sensor (FSR-402), MPU6050 Gyroscope, NodeMCU (ESP8266)
• Software: Arduino IDE, Mobile App (custom-built), Cloud Database, Machine Learning (Random Forest algorithm)
• Protocols: I2C (for gyroscope), Analog/Digital IO, Wi-Fi (ESP8266)