Signal processing and APIs for brain-computer interfaces
The MachineDigit platform is the software layer between the wearable and your application. It captures EEG brain-signal data, processes it into structured outputs, and exposes them through platform APIs and a beta dashboard. All of it is beta software: versioned, documented, and improving with every tester session.
Four layers, one platform
The MachineDigit platform decodes EEG into real-time states, a focus and attention index, per-band power, and signal quality, and exposes them over a WebSocket stream and a REST API.
Session recording from the wearable: 256 Hz, timestamped and channel-labeled EEG with device diagnostics, stored per session with tester consent.
Filtering, artifact handling, and decoding that turn raw EEG into real-time outputs: a focus and attention index, per-band power (delta, theta, alpha, beta), and per-channel signal quality. Targeting under 200 ms from signal to state.
A WebSocket stream for live state and a REST API for sessions and history, both JSON. SDKs in Python and JavaScript/TypeScript, with Unity/C# on the roadmap. Raw per-session EEG exports as EDF, CSV, or JSON.
Where testers review sessions, inspect signal quality, manage their data, and file structured feedback directly to the product team.
Which features are reliable, and which are experimental
Decoding accuracy varies by person, environment, and task. The focus and attention index, band power, and signal quality are reliable today. Drowsiness and discrete command detection are in the platform, labeled experimental, because they are. Beta testers always know which is which, and their sessions are what move features from experimental to validated.
Signal quality and filtering run on the device; decoding runs in the MachineDigit platform. Raw session EEG is uploaded under your per-session consent, encrypted in transit, and never sold.