2 out of 3 example datasets have faulty BIDS channels

[Kallemakela]

Describe the bug

Example datasets block_speech_noise and audio_or_visual_speech on this page have faulty channel order in their BIDS files. This causes a mismatch between channel names and the data when read with mne-bids (as is done on the example docs page). mne.io.read_raw_snirf seems to read the channels correctly.

Steps to reproduce

https://gist.github.com/Kallemakela/db32ddd7b96d39997047c930c7e1901d

Expected results

• Channel pairs should have the same location. E.g. S1_D1 760 and S1_D1 850 should be in the same location.
• BIDS and .snirf channels should be the same after reading (maybe this is a mne-bids issue?).

Actual results

• Channel pairs are not in the same location.
• BIDS and .snirf channels are not the same after reading.

Additional information

mne.sys_info():

Platform             macOS-14.4.1-arm64-arm-64bit
Python               3.12.12 | packaged by Anaconda, Inc. | (main, Oct 21 2025, 20:07:49) [Clang 20.1.8 ]
Executable           /Users/kallemakela/mambaforge/envs/bidstest/bin/python
CPU                  Apple M1 Pro (8 cores)
Memory               32.0 GiB

Core
├☑ mne               1.11.0.dev94+g65a717ced (development, latest release is 1.10.2)
├☑ numpy             2.3.4 (unknown linalg bindings)
├☑ scipy             1.16.3
└☑ matplotlib        3.10.7 (backend=macosx)

Numerical (optional)
├☑ sklearn           1.7.2
├☑ nibabel           5.3.2
├☑ nilearn           0.12.1
├☑ pandas            2.3.3
├☑ h5io              0.2.5
├☑ h5py              3.15.1
└☐ unavailable       numba, dipy, openmeeg, cupy

Visualization (optional)
└☐ unavailable       pyvista, pyvistaqt, vtk, qtpy, ipympl, pyqtgraph, mne-qt-browser, ipywidgets, trame_client, trame_server, trame_vtk, trame_vuetify

Ecosystem (optional)
├☑ mne-bids          0.1.0.dev1582+gf61e0e049
├☑ mne-nirs          0.8.0.dev4+gddcedfe85
├☑ eeglabio          0.1.2
├☑ edfio             0.4.10
├☑ pybv              0.7.6
├☑ defusedxml        0.7.1
└☐ unavailable       mne-features, mne-connectivity, mne-icalabel, mne-bids-pipeline, neo, mffpy, antio

[larsoner]

I think those datasets were created by @rob-luke , so it would probably need to be him that updated them. Or @rob-luke you could give me permissions on them and I could maybe try to update them

If he can't do it or doesn't respond, then maybe we could upload a separate version of the dataset somewhere else

[Kallemakela]

Here's a script for overwriting the bids dataset from .snirf files:
https://gist.github.com/Kallemakela/4cee95a58d5f176842bc9ca800769be6

I also noticed that there was a JSON syntax error in audio-visual-speech that causes an error when trying to overwrite. Fix:

bids_root="$HOME/mne_data/fNIRS-audio-visual-speech"
sed -E -i '' 's/"Colette M\. McKay",/"Colette M. McKay"/' $bids_root/dataset_description.json # Fix dataset_description.json