Brain-computer interfaces (BCIs) have become a major focus of technological innovation in recent months, with China leading the commercialization of the emerging technology. However, neighboring Australia may soon make its own mark, as the University of Melbourne startup Fluent develops a new device designed to transform how people with speech impairments communicate.
Many communication technologies currently available remain limited for people living with neurological conditions such as motor neuron disease and multiple sclerosis. As these conditions progress, many patients experience increasing difficulty speaking and may eventually lose the ability to communicate verbally altogether.
Fluent is developing a different approach from many existing BCI systems. While invasive BCIs require brain surgery to implant electrodes inside the skull, Fluent’s device is designed to sit beneath the scalp but outside the skull, offering what the company hopes will be a safer and more accessible alternative.
“Until now, this technological capability was only thought possible using highly invasive electrodes implanted inside the skull,” said Fluent co-founder and biomedical engineer Dr. Tim Mahoney in a statement.
Mahoney’s research suggests that signals captured by the device are comparable whether electrodes are placed beneath or outside the skull. If validated through further testing, this could represent a significant advance in BCI development, as invasive systems have historically captured higher-quality brain signals and outperformed non-invasive approaches.
“Our device will be positioned above an area of the brain called the motor cortex, which controls speech muscles. If you think of electrical signals as QR codes: when a person speaks, every individual mouth and jaw movement produces different ‘QR codes’ in their motor cortex,” said Mahoney.
“These ‘QR codes’ also occur when a person with impaired speech is attempting to speak. Our device can capture these codes in a sequence, which tells us what someone is trying to say,” Mahoney adds.
The device captures brain signals and records them in sequence, allowing researchers to understand what a person is attempting to communicate. A machine learning system is being developed to translate brain activity into written text or audio by just thinking about speaking.
“After building the largest English dataset of its kind, we partnered with a Japanese team leveraging an even larger dataset,” Mahoney explained. “This collaboration proved that a model can accurately isolate the correct phrase from a pool of 128 options with 96 percent accuracy.”
To date, the technology has undergone preliminary human testing at the Aikenhead Centre for Medical Discovery at St Vincent’s Hospital in Melbourne. During those trials, participants wore a cap fitted with 144 electrodes positioned over the motor cortex while speaking, miming words, and imagining different phrases.
The team’s work resulted in one of the largest English-language brain activity datasets assembled to date. In collaboration with Japanese researchers using an even larger dataset, they demonstrated that an AI model could correctly identify a target phrase from a pool of 128 possibilities with 96% accuracy.
Like other advanced medical technologies, BCIs must undergo rigorous regulatory review before they can be brought to market, and Fluent expects Australia’s approval process to follow internationally recognized standards.
“BCI is a relatively new technology, with the first devices currently being approved in China and the US. The regulatory process for BCI technology follows the same safety and efficacy data requirements used for medical device approval across the various countries’ regulatory bodies. Australia is a fantastic place to perform clinical trials due to its R&D tax incentives, meaning Fluent is well-positioned to streamline our pathway to regulatory clearance,” Mahoney said in an email to The Debrief.
With China and the United States both advancing minimally invasive BCI technologies toward commercialization, Fluent hopes its approach can distinguish itself by prioritizing accessibility over maximum signal fidelity.
“The field of BCI has made enormous progress in the past 5 years. Neuralink is close to commercializing cursor control, research groups have demonstrated online speech decoding with electrodes placed directly on the brain, and META just released their Brain2Qwerty model that can convert non-invasive ‘typing’ brain activity into text,” says Mahoney.
“The subscalp speech BCI that Fluent is building will not have the same performance as more invasive devices,” he adds. “However, we believe our solution is far more accessible, given the low-risk, set-and-forget nature of the device.”
“We also believe that our low-risk approach can deliver near-perfect speech decoding performance when paired with context-aware large language models to correct errors in an informed way,” Mahoney said.
Chrissy Newton is a PR professional and the founder of VOCAB Communications. She currently appears on The Discovery Channel and Max and hosts the Rebelliously Curious podcast, which can be found on YouTube and on all audio podcast streaming platforms. Follow her on X: @ChrissyNewton, Instagram: @BeingChrissyNewton, and chrissynewton.com. To contact Chrissy with a story, please email chrissy @ thedebrief.org.