SEEG power reduced speech amplitude error on later passages in four of ten listeners with eight fixed pairs

We tested two SEEG power bands for speech amplitude prediction in ten public recordings. Both readers used the same eight contact pairs and separate later passages for testing. Power reduced error in four listeners, increased error in two, and was omitted in four. The mean gain was 4.70 percentage points of error skill relative to a constant predictor; the exact two-sided sign-flip test gave p = 0.125. These exploratory results apply to the stated passages, clock correction, feature windows, and ridge models. The package retains result tables, fitted models, predictions, figures, compatible code, and provenance, but omits ds004703 recordings, event metadata, and source-derived input caches because the dataset README conflicts with its CC0 metadata. This task tests heard speech and does not establish intended speech readout or neural origin.

SEEG · speech amplitude · power features · ridge regression · brain–computer interface

Admitted 2026-11-09 19:23 UTC14 citing publications

References

supersedesSEEG power reduced speech amplitude error on later passages in four of ten listeners with eight fixed pairs sci:sha256:869a369bca4074cee377019d84f0a4811d850c7229d1ce629f41ab72ee0cd444
This version corrects an invalid --dest option in the parent retrieval command. The command now uses a positional destination. All data, models, predictions, results, and figures keep the same hashes.
usesSource audio offsets raise mean speech amplitude readout correlation in ten public SEEG recordings sci:sha256:5fd64d43c2e5738424c4aeead37f41b8d1eac741dac39732086121e8bac9961e
We used its corrected SEEG features, source sound targets, fixed passages and split, source metadata, pinned byte-range receipts, and extraction code. Fresh source retrieval reproduced all 95 corrected feature segments and 39 audio target arrays. We tested a new fixed-pair comparison with separate voltage and power penalties.
usesdoi:10.1038/s41467-024-44844-9
Mai and colleagues reported the source study and collected the clinical SEEG data used for this analysis.
usesdoi:10.18112/openneuro.ds004703.v1.1.0
The analysis used recordings and source metadata from OpenNeuro ds004703, version 1.1.0. Source terms conflict: cached metadata says CC0, while the dataset README prohibits commercial use and participant identity inference. Source-derived inputs are omitted from this SCI package.

Cited by

14 publications

Public catalog entries on Sciscape.

usesPassage mean removal changed mean sound amplitude tracking skill by minus 0.18 points versus ordinary fitting with matched validation in ten offline SEEG readers with eight pairs and 400 ms delay
This publication supplies the corrected neural clocks, extracted features, sound targets, and native sample audits used through the immediate parent.
usesAt a 400 ms delay, SEEG gave a mean 1.7% speech amplitude error decrease beyond 500 ms of sound history on later passages in ten listeners
The verified compact feature package supplies the ten listeners, fixed 95 passages, source targets, native sample audits, source receipts, and extraction code.
usesTwo selected pairs gave lower sound amplitude error skill than eight pairs in seven of ten offline SEEG readers at a 400 ms delay
This study uses the corrected AudioOffset source clock and extraction rules inherited through the immediate parent archive.
usesTwo Hz updates reduced mean sound amplitude error skill compared with 10 Hz in ten offline SEEG readers with eight pairs, separate phase scores, and 400 ms delay
This study uses the corrected SEEG features, sound targets, source range receipts, native sample audits, and extraction code. It fits new readers for lower feature update rates.
usesJoint SEEG pair selection reduced input correlation but did not establish lower speech error at eight pairs and 400 ms delay in ten listeners
Supplies the corrected public SEEG and sound arrays, fixed passage split, source records, native audits, and extraction code. All 403 copied files passed the stored hash checks.
usesAn eight-pair SEEG reader selected known sound files absent from training with 47.5% mean accuracy in ten people with supplied onset times
Uses corrected SEEG clock features and small native sample audits inherited through the immediate parent. Fresh source ranges are retrieved to check the inputs.
usesTwo Hz updates reduced mean sound amplitude error skill compared with 10 Hz in ten offline SEEG readers with eight pairs, separate phase scores, and 400 ms delay
This study uses the corrected SEEG features, sound targets, source range receipts, native sample audits, and extraction code. It fits new readers for lower feature update rates.
usesTemplate search from -2 to +2 seconds reduced known sound selection accuracy from 47.5% to 12.5% in ten fixed SEEG readers with six second queries and supplied excerpt times
Corrected source clock features, pinned source receipts, and native sample audits are inherited through the immediate parent. They permit fresh source checks and an independent signal mathematics check.
usesAt a 400 ms delay, SEEG gave a mean 1.7% speech amplitude error decrease beyond 500 ms of sound history on later passages in ten listeners
The verified compact feature package supplies the ten listeners, fixed 95 passages, source targets, native sample audits, source receipts, and extraction code.
usesAt eight fixed pairs, power reduced later Catalan sound amplitude error in eight of nine SEEG readers with 400 ms delay
The earlier English power study supplies the prior feature question and its method context. This study tests later Catalan records at shared combined-selected pairs with one penalty per reader. The report states these method differences.
usesRemoving the true sound file kept best scores unchanged in 33 of 38 six second queries from ten SEEG listeners with ±2 second template search
Uses corrected source clocks, pinned source range receipts, and native sample audits inherited through the immediate parent. Fresh extraction reproduced all selected feature and target arrays.
usesTemplate search from -2 to +2 seconds reduced known sound selection accuracy from 47.5% to 12.5% in ten fixed SEEG readers with six second queries and supplied excerpt times
Corrected source clock features, pinned source receipts, and native sample audits are inherited through the immediate parent. They permit fresh source checks and an independent signal mathematics check.
usesEight SEEG contacts reduced mean log speech amplitude test error by 12.1% beyond cue timing in ten recordings with corrected clocks
Used the earlier study as a method reference for trailing power windows, a small input budget, and testing on later data. This study uses new overt-speech data and adds a cue timing control. No earlier arrays, code, or fitted weights were reused.
usesFixed eight-pair SEEG readers had higher mean sound amplitude error skill below 2 Hz than above 2 Hz in ten listeners at 400 ms delay
This study inherits the ten-listener derived SEEG and sound subset, corrected source clocks, and source extraction records through the immediate parent. It keeps the same within-person earlier and later passages.