At 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
Public SEEG from ten listeners was used to reconstruct heard speech amplitude. All readers were fitted on earlier passages. A sound history reader used 500 ms of known preceding sound. The combined reader could add eight SEEG pairs at a 400 ms delay. The mean relative test error decrease was 1.70%. The listener bootstrap 95% interval was -3.19% to 6.46%. Five listeners improved, three worsened, and two omitted SEEG. These data do not show a clear mean benefit under this model. The package includes source features, fitted models, figures, source receipts, and code. This is an exploratory test on heard sound. It does not establish intended speech readout or neural signal origin.
SEEG · speech amplitude · brain-computer interface · ridge regression · sound history · electrophysiology
References
5- supersedesAt 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 sci:sha256:8445007c411326e90836eb8eac207c50834888187a80a886f055b730c947beb2
The README corrects the minimum Python version to 3.11 and states the environment assumption for exact replay. Scientific data, results, models, predictions, and figures are unchanged.
- usesSEEG power reduced speech amplitude error on later passages in four of ten listeners with eight fixed pairs sci:sha256:9d7986c2893a49c9a3d06daa806e2fe9a866569d94a0968c0542dfed56513cc9
The verified compact feature package supplies the ten listeners, fixed 95 passages, source targets, native sample audits, source receipts, and extraction code.
- usesSource audio offsets raise mean speech amplitude readout correlation in ten public SEEG recordings sci:sha256:5fd64d43c2e5738424c4aeead37f41b8d1eac741dac39732086121e8bac9961e
The clock correction and source range extraction were retained through the immediate input package. The source session AudioOffset defines alignment.
- usesdoi:10.18112/openneuro.ds004703.v1.1.0
The study used source ranges from OpenNeuro ds004703. Version-pinned records and hashes identify the analyzed source files, although they do not establish that each live S3 object is byte-identical to the DOI snapshot.
- doi:10.1038/s41467-024-44844-9
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesPast SEEG ridge readers did not lower 100 ms sound forecast error beyond quadratic sound history in ten listeners
This study uses the corrected compact SEEG data, sound targets, source range receipts, native sample audits, and extraction code. It adapts the history-control ridge code and tests a shared input cutoff for future sound forecasts.