Building Signal2Speech: answering questions with an EEG headset at Hack the Nest 2026 (Update 1)
Note: This is only the first of many updates. We will also include a final post which will summarize the whole event. We are Order of the Golden Eagles, a team of four at Hack the Nest 2026 (Eastern Michigan University), and we are building Signal2Speech. It is an accessibility prototype for people who have difficulty speaking or typing. A conversation partner asks a question, the app offers a…
Order of the Golden Eagles, a team of four, is developing Signal2Speech, an accessibility prototype for people with difficulty speaking or typing. The app presents a few short replies to a conversation partner's question, and the wearer wearing an EEG headset selects, previews, and confirms the desired words. The team is currently five and a half hours into the project.
The selection process involves a P300 chooser, which detects a brain response called the P300 that appears about 300 milliseconds after a stimulus flashes. A Linear Discriminant Analysis (LDA) algorithm classifies the stimuli. The wearer then confirms the chosen reply by closing their eyes for about two seconds, which raises the brain's alpha rhythm. This alpha switch converts the confirmation into a single "Confirm" signal, after which the app speaks the chosen words.
The project has faced several challenges. Initially, the team planned to build the app in Unity with g.tec's starter kit. However, they switched to Python, g.Pype (g.tec's Python toolkit for EEG), and PySide6 due to the benefits of g.Pype in visualizing the signal pipeline, including OSCAR artifact removal. The existing selection rules remained unchanged as they were written as shared scenario files.
The current functionality includes the full choose, preview, and confirm flow, running end-to-end on keyboard input and simulated brain data. OpenOrbit, a feature allowing the wearer to explore Earth data, has real data bundled in the repository for a fully functional demo. The team added a NASA FIRMS map to show the smoke's source, which turned out to be from Quebec and Ontario, rather than Michigan.
A bug was encountered when the P300 chooser and the alpha switch made their own live EEG adapters, resulting in conflicting event IDs. To resolve this, a shared counter was added to ensure unique IDs for all adapters. The team also implemented safeguards to prevent accidental or late/similar signals from being spoken, ensuring the wearer's intent is respected.
The next steps involve connecting the g.tec Unicorn Hybrid Black to the app, calibrating it on an actual wearer, and running the complete flow with the real headset. Once complete, they plan to record a demo video.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
