Add Sherlock 9 stimulus README

Sherlock9/stimuli/README.md

@@ -0,0 +1,85 @@
+# The Case-Book of Sherlock Holmes audio stimuli
+
+This directory contains the LibriVox audiobook audio used as auditory stimuli
+for Sherlock 9 in LibriBrain100.
+
+## Source
+
+The stimulus material is *The Case-Book of Sherlock Holmes* by Arthur Conan
+Doyle, read by Thomas A. Copeland for LibriVox.
+
+LibriVox source:
+
+https://librivox.org/the-case-book-of-sherlock-holmes-by-sir-arthur-conan-doyle/
+
+The corresponding text was obtained from Project Gutenberg.
+
+Both the text and the LibriVox recording used here are public-domain material.
+
+## Files
+
+The `audio/` directory contains the 13 WAV files used in the LibriBrain100
+MEG experiment:
+
+- `casebookofsherlockholmes_00_doyle_64kb.wav`
+- `casebookofsherlockholmes_01_doyle_64kb.wav`
+- ...
+- `casebookofsherlockholmes_12_doyle_64kb.wav`
+
+Each file corresponds to one MEG session / audiobook chapter.
+
+The total stimulus duration is approximately 8 h 27 min.
+
+## Stimulus preparation
+
+For the LibriBrain100 experiment, the source audiobook audio was converted to
+uncompressed WAV format and resampled to 48 kHz using SoX.
+
+The audio was segmented into natural sentences or phrases using voice
+activity detection followed by manual review and correction.
+
+The corresponding text was obtained from Project Gutenberg and then manually
+corrected to match the spoken audiobook. Text was also normalised to reflect
+the spoken form, for example by spelling out numbers and resolving other
+written forms that were ambiguous or differed from the narration.
+
+## Linguistic annotations
+
+Time-aligned linguistic annotations are provided in the `*_events.tsv` files
+alongside the corresponding MEG recordings in the dataset.
+
+These include word- and phoneme-level annotations aligned to the stimulus
+audio.
+
+## Relationship to the MEG data
+
+Each audio file corresponds to a Sherlock 9 MEG session in the BIDS dataset.
+
+The stimulus audio is provided here to make it possible to reconstruct the
+auditory input presented during the experiment.
+
+## Dataset split
+
+All Sherlock 9 sessions are part of the training data in the LibriBrain100
+benchmark.
+
+Validation and test data are drawn from held-out sessions from *A Study in
+Scarlet*, rather than from Sherlock 9.
+
+## Licence
+
+The audiobook recording and source text used here are public-domain
+materials.
+
+The LibriBrain100-specific annotations and dataset organisation are released
+under the applicable LibriBrain100 dataset terms.
+
+## Citation
+
+If you use these stimuli together with the corresponding MEG data, please
+cite the LibriBrain100 dataset paper.
+
+Please also credit the original LibriVox recording and reader:
+
+Arthur Conan Doyle, *The Case-Book of Sherlock Holmes*.
+Read by Thomas A. Copeland. LibriVox.