A 60 dB FIR design kept 500 ms, 70–108 Hz SEEG power within 5% after 4096-to-256 Hz reduction in 18 people

We tested two fixed FIR filter designs on public SEEG from OpenNeuro ds006519 version 1.0.0. The main group had 31 records from 18 people at 4096 Hz. Power used 500 ms frames and the 2 Hz bins from 70 through 108 Hz. At 256 Hz, the 60 dB design kept every scored power value within 5% of its native-rate reference. Mean person error was 0.001329 dB. The 80 dB design had error 0.000110 dB. The shorter design reduced added filter delay from 140.625 to 101.5625 ms, a reduction of 27.78%. At 512 Hz, delay fell by 22.22% with mean error 0.003895 dB. Four native 512 Hz records from three more people were reported separately. One zero-power channel was excluded under a fixed rule. Independent raw-channel checks passed. A full source rebuild matched 92 numerical outputs. The package preserves all scored powers, raw audit channels, source hashes, code, figures, plans, and reproduction steps. The result measures recorded power fidelity. It does not establish decoder accuracy or artifact-free neural signals.

stereoelectroencephalography · SEEG · signal processing · FIR filter · sample-rate reduction · power fidelity · group delay · brain-computer interface