Tether Evo has three new peer-reviewed papers on Brain-Computer Interfaces (BCIs) accepted for publication at the Journal of Neural Engineering, Imaging Neuroscience, and Neural Networks.
Two of the papers were developed by Tether Evo in collaboration with the University of Rome Tor Vergata (UniTOV).
The research conducted across three papers shows how a single model can learn to read the brain activity across different people, instead of starting from scratch with every patient. They do so in three ways: speech, vision, and music.
This addresses one of the biggest challenges in brain-computer interface (BCI) research, given that every brain produces slightly different signals.
One model, many brains
Tether Evo is Tether’s frontier technology division, dedicated to the intersection of biology and machine intelligence. Specialising in BCI and neuroprosthetics, Tether Evo builds local-first, high-performance systems designed to empower individuals and preserve personal autonomy in an increasingly centralised world.
Although every brain is different, and across different people representations of speech can vary (due to different implantation location, different functional activity, and different learning curricula), researchers have been able to show that a single model, trained with the right alignment technique, can generalise across people rather than being rebuilt for each one.
In the paper, Cross-subject decoding of human neural data for speech brain computer interfaces, accepted at the Journal of Neural Engineering, researchers at Tether Evo explore how BCIs can help people who’ve lost the ability to speak, due to ALS, stroke, or brain injury, helping them communicate again, by reading brain activity and decoding it into text.
Until now, these systems had to be built separately for each patient, since every brain’s signals are different. The paper proposes the first cross-subject neural-to-phoneme decoding model trained on invasive recordings from multiple participants implanted in distinct cortical regions.
Rather than treating each participant as an isolated case, Tether’s approach leverages shared speech patterns. It uses this shared understanding to train one decoder that can then be quickly adjusted for each new patient.
This study shows that a single model can instead be trained across multiple people at once, using a lightweight mathematical realignment step to bring different brains’ signals into a shared space, along with a new layered decoding network. The result matches or beats today’s single-patient systems, while adapting to a new person in minutes to hours instead of the usual lengthy calibration, a promising step toward faster, more scalable speech BCIs.
BCI for vision reconstruction
Researchers from Tether and UniTOV collaborated to record brain signals from macaques viewing thousands of images and reconstructed what the animals were seeing directly from that activity. From 200 milliseconds of neural data, the model picked the exact image out of thousands with 70% accuracy, and generated a plausible reconstruction capturing its shape, colour, and content.
A Modular Semantic-Structural Pipeline for Visual Decoding from Primate Spiking Data via Selective Temporal Integration, accepted at Imaging Neuroscience, showcases how BCI systems can support visual reconstruction, with potential applications for patients with vision loss by advancing visual restoration research for future cortical visual prostheses and closed-loop BCIs.
The role of BCI in music
The same alignment approach was demonstrated in the co-authored paper R&B – rhythm and brain: Cross subject decoding of music from human brain activity, accepted at Neural Networks. In this study, researchers recorded brain scans (fMRI) from five people as they listened to 540 songs spanning 10 genres, then built a model that translates neural activity into an AI representation of the music itself.
Using a technique that aligns brain patterns across different people’s brains, the model could identify the correct genre being heard about 61% of the time (versus 10% by chance), and pinpoint the exact song among 60 candidates roughly 25% of the time (versus under 2% by chance), setting a new benchmark for this task.
The results also mapped out which brain regions are doing the work, pointing to auditory areas long linked to music perception, demonstrating that classical and jazz produce the most distinctive brain signatures. Genres like metal and disco are more easily confused.
What this means for future BCI
Calibration time has been one of the biggest barriers to real-world BCI adoption. If a model trained on many people can adapt to someone new in minutes rather than weeks, the timeline from lab result to usable device gets shorter. The same logic that lets a system recognise a song or an image applies directly to restoring communication for someone who has lost it.
Tether believes that nothing is more private than your own information. It has built QVAC, Tether’s open-source on-device AI stack, to ensure that intelligence runs privately, locally, and without permission on any device. This aligns with Tether’s broader aim to empower individuals and communities through decentralised technologies, wallets, and tools that are entirely self-owned.
