AI Smart Meditation Headband based on TGAM

Client

A mindfulness technology company creating an affordable consumer EEG headband for guided meditation with real-time brainwave feedback.

Challenge

Extracting clean alpha/theta brainwave ratios from a low-cost dry-electrode EEG sensor in a noisy home environment, with motion artifact rejection.

Product Concept & Architecture

We used the TGAM (ThinkGear AM) EEG sensor module with three dry silver electrodes (FP1, FP2, reference). The Nordic nRF52840 processes raw brainwave data and streams compressed spectrograms to a smartphone via BLE 5.0. Bone conduction audio provides guided meditation without blocking ambient awareness.

Engineering Solution

The nRF52840 runs a real-time FFT pipeline on EEG data at 512Hz. Artifact rejection uses a cascade of threshold and statistical filters: eye-blink spikes (>150µV), muscle tension (>100µV gamma), and movement artifacts are blanked. A relaxation score is computed from the alpha/theta ratio. When the ratio indicates deep relaxation, the bone conduction guidance gradually fades to silence.

Result

EEG signal quality comparable to research-grade wet-electrode systems (r²=0.89). 92% of users reached a meditative state within 5 minutes in guided mode.