In one SEEG patient, 70–150 Hz sentence responses did not identify talker sex, and no male-F0 band difference was detected

High-frequency activity (HFA; 70–150 Hz power) is commonly used as a proxy for local cortical activation in speech studies. Male voice fundamental frequency (F0; approximately 85–155 Hz) overlaps this band, whereas female voice F0 (approximately 165–255 Hz) is above it. In one patient with stereo-EEG (SEEG), we tested whether sentence-level HFA identified TIMIT talker sex when entire talkers were held out, and whether the male–female difference was greater in 90–150 Hz than in 170–250 Hz. The analyzed recording was OpenNeuro ds004993, snapshot 1.1.2, sub-W3, task-timit4, run-01: 126 sentences (92 male, 34 female) from 122 talkers and 148 same-shaft bipolar pairs. A label-blind test identified 39 responsive pairs. Talker-grouped 10-fold cross-validation gave balanced accuracy 0.567 (talker-level permutation p = 0.136; null 95th percentile 0.599). The F0-band contrast was +0.0016 log10 units (p = 0.84; Holm p = 0.84). With sentence duration regressed out within training folds, balanced accuracy was 0.576 (p = 0.104; Holm p = 0.21). No 10-Hz band from 70 to 250 Hz had a per-band p below 0.12. Thus, these data did not show talker-sex information in held-out talkers or a male-F0 component in HFA for this patient and these contacts. With 34 female sentences, effects below approximately 0.60 balanced accuracy were not detectable by this design. There was one patient and one run; audio was unavailable, so sentence loudness and F0 were not measured, and a reported VNS artifact was not removed.

stereo-EEG · high-frequency activity · speech · talker sex · TIMIT · fundamental frequency · cross-validation · intracranial electrophysiology