In OpenNeuro ds004703 SEEG, speech-envelope tracking occurs in 7 of 9 subjects at +0.01 to +0.19 s

We tested whether high-gamma (70–150 Hz) activity in stereo-EEG tracks natural English speech envelopes under the alignment rule t_EEG = t_log + AudioOffset. Across the first four English passages in nine subjects, a cross-validated test found tracking in 7/9 subjects (held-out r = 0.07–0.34, permutation p ≤ 0.002; binomial p = 2.6 × 10⁻⁸ against 0.05). Peak lags in those subjects were +0.01 to +0.19 s, with neural activity after sound, supporting the alignment rule; no residual offset larger than about 0.2 s was found. Best pairs were mostly on temporal shafts, based on contact names only. A check of an earlier null result found that its 10-sample bins at 1024 Hz (9.77 ms) were treated as 10 ms bins, causing a 2.4% time drift. Applying this binning reduced peak correlations from 0.12–0.35 to 0.02–0.14 and shifted peak lags to −2.4 to +0.8 s. SD018 was not included in the reported analysis after repeated connection resets and a time limit. Only four passages per subject were analysed; the lag estimate uses the best pair selected on the same data, while cross-validation controls the tracking test, not the lag estimate. Low-pass filtering and possible sound-output latency limit resolution below about 0.05 s. Dataset files were read from the live OpenNeuro mirror; the exact release is not recorded.

stereo-EEG · speech envelope · high gamma · auditory cortex · cross-correlation · signal alignment · time base