In a promising leap forward for assistive communication technologies, Jihwan Lee, a leading researcher at the University of Southern California (USC), is developing non-invasive methods to decode speech directly from brain activity. His work could dramatically improve the lives of people with speech impairments due to conditions like ALS, stroke, or paralysis.
Moving Beyond Invasive Methods
Until recently, decoding complex brain functions such as speech required invasive procedures like electrocorticography (ECoG), where electrodes are surgically implanted into the brain. Lee and his team at USC’s Ming Hsieh Department of Electrical and Computer Engineering and the Signal Analysis and Interpretation Laboratory (SAIL) are breaking new ground by using electroencephalography (EEG)—a non-invasive, wearable approach.
“Interpreting brain signals with non-invasive methods like EEG was traditionally limited to simple tasks,” said Lee. “But recent research, including ours, shows that even complex functions like speech decoding are now within reach.”
A Wearable Future for Brain-Computer Interfaces
The impact of this research could be profound. A wearable EEG cap could one day allow users to convert thoughts into speech without moving a muscle. This non-invasive approach makes brain-computer interfaces (BCIs) more accessible and safer, eliminating the need for surgery.
Thanks to rapid advances in artificial intelligence and signal processing, the system can now interpret a user’s intent to speak—or even partial movements of speech muscles—and translate those signals into audible or textual speech.
Tackling the Challenge of Accuracy
Despite the exciting progress, Lee acknowledged that high-accuracy speech decoding still largely depends on invasive methods. “That’s a big barrier for those who can’t or won’t undergo surgery. Our goal is to replicate these capabilities using EEG instead,” he explained. “While we’ve made progress, these non-invasive systems are still in the early stages and not yet ready for widespread use.”
Power of Global Collaboration
Lee emphasized the value of international partnerships in advancing the field. His team collaborates with institutions like IIT-Madras and the Indian Institute of Science to study diverse populations and adapt to varied healthcare guidelines.
A key moment in this collaboration came during a recent workshop in Hyderabad, India, focused on brain-body-behavior dynamics in human speech. “This event helped strengthen our global research dialogue,” said Lee.
Breakthrough Parallel Architecture for Real-Time Decoding
What truly distinguishes USC’s research is its novel dual-module system that processes EEG brain signals into speech in real-time. Instead of following a traditional, step-by-step decoding method, this new architecture runs two modules—one predicting phoneme sequences and the other generating waveforms—simultaneously.
“This parallel structure reduces latency and prevents errors from building up, improving overall performance,” said Lee. The system delivers more natural-sounding speech and increases the accuracy of textual interpretations, making it far more practical for real-world use.
Enhanced Model Doubles Previous Performance
The research team’s latest paper, Enhancing Listened Speech Decoding from EEG via Parallel Phoneme Sequence Prediction, showcases this advanced system. According to Shrikanth Narayanan, USC University Professor and Director of SAIL, the new model is twice as effective as its predecessor from last year.
“This tool significantly reduces noise and errors in EEG-based speech decoding,” Narayanan explained. “The convergence of breakthroughs in biosensing, AI, and signal processing is giving us powerful new insights into the brain-body-behavior link. These technologies will soon support and enhance human communication in ways we couldn’t imagine before.”
A New Era for Communication Technology
With continued research and collaboration, non-invasive speech BCIs may soon offer real-time, accurate, and effortless communication for individuals who cannot speak. As reported by pharmabiz.com, Lee and his team’s work at USC is lighting the path toward that future—where thoughts can truly become words, no surgery required.