RU NEWS
Brain2Frame AI Student's Project
Giving Thoughts a Voice: Raffles University AI Student Develops Brain2Frame to Empower Accessible Communication
For individuals living with severe speech or mobility impairments, expressing even the simplest everyday request—such as asking for a bottle of water—can require immense physical effort or specialised medical hardware. At Raffles University, Bachelor of Information Systems in Artificial Intelligence (Honours) student Tharsshen A/L Govikrishnan is tackling this challenge head-on through Brain2Frame, an innovative brain-computer interface (BCI) project designed to translate human thoughts directly into real-time visuals.
Bridging the Communication Gap Without the Lab
Most traditional brain-reading research relies on costly, hospital-grade equipment—caps fitted with 64 to 128 electrodes, conductive gel, and complex laboratory setups. Brain2Frame reimagines that approach by pairing artificial intelligence with a lightweight, consumer-friendly Muse 2 EEG headband equipped with just four scalp sensors. Affordable, portable, and completely non-invasive, the headband streams electrical brain activity wirelessly, allowing users to interact with a computer simply by thinking—eliminating the need for typing, speaking, or physical movement.
How Brain2Frame Works
Because every individual’s neural patterns are unique, Brain2Frame is built to adapt to each wearer on the fly without requiring a model to be retrained from scratch. Before a live session begins, the user undergoes a quick personal calibration by thinking through a short set of baseline sentences (“A red car drives down the highway”, “She plays piano in the evening”, and “A child laughs in the playground”), allowing the system to observe their brainwave activity and build a personalised neural fingerprint.
During live operation, rather than rigidly forcing signals into fixed categories, the system reads the directional patterns of the user’s brain activity to determine which underlying concept—such as car, piano, or child—their thoughts align with most closely, displaying its prediction alongside a confidence score. Once a sentence is predicted, Brain2Frame sends it to Google’s Veo3 video-generation AI, which renders a four-second cinematic clip of that exact scene within seconds—turning an unspoken thought into a visual message on screen.
Technology With Real Human Impact
By proving that a $250 consumer headset can reliably distinguish between distinct thoughts in real time, Brain2Frame highlights the growing potential of accessible assistive technology. Projects like Brain2Frame reflect Raffles University’s commitment to nurturing future-ready innovators who pair strong technical foundations in artificial intelligence with empathy and real-world problem-solving. Through initiatives that prioritise accessibility and low barriers to entry, students are demonstrating how emerging technologies can make everyday communication more inclusive for everyone.