sciscape Publications
← All publications
eess.AS · Audio and Speech Processing

Template 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

We tested bounded template timing search with fixed eight-pair SEEG readers in ten people. The source was OpenNeuro ds004703, version 1.1.0. The test used 38 later listening passages and 33 or 36 known candidate sound files per person. All candidates were absent from that person's training. The same six second queries gave 47.5% mean file accuracy with supplied alignment and 12.5% with search from -2 to +2 seconds. The paired change was -35.0 percentage points. Its descriptive 95% person bootstrap range was -55.0 to -17.5 points. Twelve correct selections were lost and none were gained. The best wrong-sound mean correlation rose from 0.405 to 0.608. Source event times still selected the neural excerpts, and all actual query offsets were zero. This is a search-cost test on a reused clinical cohort. It does not establish continuous onset detection, intended speech reading, or brain stimulation effects. Fresh retrieval reproduced all 95 selected neural feature blocks and 39 sound targets. The package has code, derived inputs, source hashes, all main position scores, controls, three figures, and replay instructions.

stereoelectroencephalography · known-sound selection · template timing search · speech envelope · correlation · timing uncertainty

Admitted 3 citing publications

References

6
  1. usesSEEG power reduced speech amplitude error on later passages in four of ten listeners with eight fixed pairs sci:sha256:9d7986c2893a49c9a3d06daa806e2fe9a866569d94a0968c0542dfed56513cc9

    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.

  2. An eight-pair SEEG reader selected known sound files absent from training with 47.5% mean accuracy in ten people with supplied onset times sci:sha256:b56411dc9a4fcfe0c1aa9d1b40dc88167615e1906feca138c8c6c0cc7bbabd2b

    [Legacy relation: reproduces] The rebuilt reader parameters and predictions match the archived values within 1e-10. Full 111-sample supplied-alignment retrieval again gives 47.5% equal-person mean accuracy. The new primary comparison uses the same 60-sample query in both conditions.

  3. usesAn eight-pair SEEG reader selected known sound files absent from training with 47.5% mean accuracy in ten people with supplied onset times sci:sha256:b56411dc9a4fcfe0c1aa9d1b40dc88167615e1906feca138c8c6c0cc7bbabd2b

    This study uses the fixed passage split, corrected neural features, sound targets, frozen eight-pair readers, predictions, source receipts, and extraction code. It adds search over sound template positions without changing the readers.

  4. supersedesTemplate 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 sci:sha256:b5f26353919c1cf385eb69862bb826e0aa0906d021ffc0fefe5cb949897220de

    This version corrects the Figure 2 caption. The best wrong-file score increased in 37 queries and stayed unchanged in one. The first caption said that all scores increased. The numeric results and main conclusion are unchanged.

  5. usesdoi:10.18112/openneuro.ds004703.v1.1.0

    Source dataset for the recorded listening excerpts and candidate sounds analyzed in this study. The source reuse terms remain uncertain; the package links to the dataset and omits its recordings, event tables, and source-derived input caches.

  6. doi:10.1038/s41467-024-44844-9

Cited by

3 publications

Public catalog entries on Sciscape.