Lorraine Underwood's Raspberry Pi-Powered Speech-to-Text Mask Tackles a Key Accessibility Issue
Designed to assist those who struggle to hear speech when someone is wearing a mask, this project is more than a fashion accessory.
Maker Lorraine Underwood has unveiled her latest project: A light-up face mask, powered by a Raspberry Pi single-board computer (SBC), which converts the wearer's speech into scrolling text to improve accessibility.
"When we wear a face mask it muffles the sound of our voice," Underwood explains of the project's inspiration, "and it also makes it really hard for people who rely on lip reading to understand what we're saying. I have seen face masks that have a transparent panel here at the front, but I've seen them kind of steam up and they also just look a bit strange."
"My idea is to create an RGB matrix of lights that goes across the front of the face mask with a microphone inside the mask, so when I speak it transcribes what I'm saying in real-time across the front of the mask — so people can read what I'm saying while also trying to keep eye contact on the face."
The core of the project is a Raspberry Pi, linked to a microphone and a battery for portable power. An RGB LED matrix is connected to the Pi's general-purpose input/output (GPIO) header and driven using Adafruit's NeoPixel library, with Python code sending text to the matrix for display — leaving only the trick of turning spoken word into visible text.
"We're looking at the library called 'speech recognition' for this," Underwood notes. "It works with a free Google API just for testing purposes — anything more than that you need to sign up for the paid API from Google, IBM, Amazon, and there's a few others. I'm going to just stick with Google, I found that one the easiest one to install. It lets you see the Python — you don't have to learn any new software."
The finished mask is taken on a trial run, and various tweaks developed along the way — including an ESP32-based variant linked via Bluetooth to a desktop PC at a doctor's surgery to allow the staff to be better understood by patients who would normally rely on lip-reading. Full details, including additional videos and a bill of materials, can be found on the element14 website.