Braindecode 1.3.0.dev169126795__tar.gz → 1.3.0.dev169765289__tar.gz

{braindecode-1.3.0.dev169126795/braindecode.egg-info → braindecode-1.3.0.dev169765289}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: braindecode
-Version: 1.3.0.dev169126795
+Version: 1.3.0.dev169765289
 Summary: Deep learning software to decode EEG, ECG or MEG signals
 Author-email: Robin Tibor Schirrmeister <robintibor@gmail.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Alexandre Gramfort <agramfort@meta.com>
 Maintainer-email: Alexandre Gramfort <agramfort@meta.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Robin Tibor Schirrmeister <robintibor@gmail.com>
@@ -14,44 +14,47 @@ Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
 Classifier: Topic :: Software Development :: Build Tools
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
-Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
-Requires-Python: >=3.10
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Requires-Python: >=3.11
 Description-Content-Type: text/x-rst
 License-File: LICENSE.txt
 License-File: NOTICE.txt
-Requires-Dist: mne>=1.10.0
+Requires-Dist: mne>=1.11.0
 Requires-Dist: mne_bids>=0.16
 Requires-Dist: numpy
 Requires-Dist: pandas
 Requires-Dist: scipy
 Requires-Dist: matplotlib
 Requires-Dist: h5py
-Requires-Dist: skorch>=1.2.0
-Requires-Dist: torch<3.0,>=2.0
-Requires-Dist: torchaudio<3.0,>=2.0
+Requires-Dist: skorch>=1.3.0
+Requires-Dist: torch>=2.0
+Requires-Dist: torchaudio>=2.0
 Requires-Dist: einops
 Requires-Dist: joblib
-Requires-Dist: torchinfo~=1.8
+Requires-Dist: torchinfo
 Requires-Dist: wfdb
-Requires-Dist: h5py
 Requires-Dist: linear_attention_transformer
 Requires-Dist: docstring_inheritance
 Requires-Dist: rotary_embedding_torch
 Provides-Extra: moabb
-Requires-Dist: moabb>=1.2.0; extra == "moabb"
+Requires-Dist: moabb>=1.4.2; extra == "moabb"
 Provides-Extra: eegprep
-Requires-Dist: eegprep[eeglabio]>=0.1.1; extra == "eegprep"
+Requires-Dist: eegprep[eeglabio]>=0.2.23; extra == "eegprep"
 Provides-Extra: hub
 Requires-Dist: huggingface_hub[torch]>=0.20.0; extra == "hub"
-Requires-Dist: zarr<3.0,>=2.18; extra == "hub"
+Requires-Dist: zarr>=3.0; extra == "hub"
 Provides-Extra: tests
 Requires-Dist: pytest; extra == "tests"
 Requires-Dist: pytest-cov; extra == "tests"
 Requires-Dist: codecov; extra == "tests"
 Requires-Dist: pytest_cases; extra == "tests"
 Requires-Dist: mypy; extra == "tests"
+Provides-Extra: typing
+Requires-Dist: exca==0.4; extra == "typing"
+Requires-Dist: numpydantic>=1.7; extra == "typing"
 Provides-Extra: docs
 Requires-Dist: sphinx_gallery; extra == "docs"
 Requires-Dist: sphinx_rtd_theme; extra == "docs"
@@ -70,14 +73,21 @@ Requires-Dist: seaborn; extra == "docs"
 Requires-Dist: pre-commit; extra == "docs"
 Requires-Dist: openneuro-py; extra == "docs"
 Requires-Dist: plotly; extra == "docs"
+Requires-Dist: shap; extra == "docs"
+Requires-Dist: nbformat; extra == "docs"
 Provides-Extra: all
 Requires-Dist: braindecode[moabb]; extra == "all"
-Requires-Dist: braindecode[hub]; extra == "all"
 Requires-Dist: braindecode[tests]; extra == "all"
 Requires-Dist: braindecode[docs]; extra == "all"
+Requires-Dist: braindecode[hub]; extra == "all"
 Requires-Dist: braindecode[eegprep]; extra == "all"
+Requires-Dist: braindecode[typing]; extra == "all"
 Dynamic: license-file
 
+.. image:: https://zenodo.org/badge/232335424.svg
+   :target: https://doi.org/10.5281/zenodo.8214376
+   :alt: DOI
+
 .. image:: https://badges.gitter.im/braindecodechat/community.svg
    :alt: Join the chat at https://gitter.im/braindecodechat/community
    :target: https://gitter.im/braindecodechat/community?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge
@@ -93,6 +103,7 @@ Dynamic: license-file
    :target: https://codecov.io/gh/braindecode/braindecode
    :alt: Code Coverage
 
+
 .. |Braindecode| image:: https://user-images.githubusercontent.com/42702466/177958779-b00628aa-9155-4c51-96d1-d8c345aff575.svg
 .. _braindecode: braindecode.org/
 
@@ -149,7 +160,15 @@ https://gitter.im/braindecodechat/community
 Citing
 ======
 
-If you use this code in a scientific publication, please cite us as:
+If you use Braindecode in scientific work, please cite the software using the Zenodo DOI shown in the badge below:
+
+.. image:: https://zenodo.org/badge/232335424.svg
+   :target: https://doi.org/10.5281/zenodo.8214376
+   :alt: DOI
+
+Additionally, we highly encourage you to cite the article that originally introduced
+the Braindecode library and has served as a foundational reference for many works
+on deep learning with EEG recordings. Please use the following reference:
 
 .. code-block:: bibtex
 

{braindecode-1.3.0.dev169126795 → braindecode-1.3.0.dev169765289}/README.rst

@@ -1,3 +1,7 @@
+.. image:: https://zenodo.org/badge/232335424.svg
+   :target: https://doi.org/10.5281/zenodo.8214376
+   :alt: DOI
+
 .. image:: https://badges.gitter.im/braindecodechat/community.svg
    :alt: Join the chat at https://gitter.im/braindecodechat/community
    :target: https://gitter.im/braindecodechat/community?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge
@@ -13,6 +17,7 @@
    :target: https://codecov.io/gh/braindecode/braindecode
    :alt: Code Coverage
 
+
 .. |Braindecode| image:: https://user-images.githubusercontent.com/42702466/177958779-b00628aa-9155-4c51-96d1-d8c345aff575.svg
 .. _braindecode: braindecode.org/
 
@@ -69,7 +74,15 @@ https://gitter.im/braindecodechat/community
 Citing
 ======
 
-If you use this code in a scientific publication, please cite us as:
+If you use Braindecode in scientific work, please cite the software using the Zenodo DOI shown in the badge below:
+
+.. image:: https://zenodo.org/badge/232335424.svg
+   :target: https://doi.org/10.5281/zenodo.8214376
+   :alt: DOI
+
+Additionally, we highly encourage you to cite the article that originally introduced
+the Braindecode library and has served as a foundational reference for many works
+on deep learning with EEG recordings. Please use the following reference:
 
 .. code-block:: bibtex
 

{braindecode-1.3.0.dev169126795 → braindecode-1.3.0.dev169765289}/braindecode/augmentation/__init__.py

@@ -5,8 +5,10 @@ Utilities for data augmentation.
 from . import functional
 from .base import AugmentedDataLoader, Compose, IdentityTransform, Transform
 from .transforms import (
+    AmplitudeScale,
     BandstopFilter,
     ChannelsDropout,
+    ChannelsReref,
     ChannelsShuffle,
     ChannelsSymmetry,
     FrequencyShift,
@@ -46,5 +48,7 @@ __all__ = [
     "Mixup",
     "SegmentationReconstruction",
     "MaskEncoding",
+    "AmplitudeScale",
+    "ChannelsReref",
     "functional",
 ]

{braindecode-1.3.0.dev169126795 → braindecode-1.3.0.dev169765289}/braindecode/classifier.py

@@ -8,7 +8,6 @@
 
 import warnings
 
-import numpy as np
 from skorch import NeuralNet
 from skorch.callbacks import EpochScoring
 from skorch.classifier import NeuralNetClassifier
@@ -237,13 +236,9 @@ class EEGClassifier(_EEGNeuralNet, NeuralNetClassifier):
             num_workers=self.get_iterator(X, training=False).loader.num_workers,
         )
 
-    def _get_n_outputs(self, y, classes):
-        classes_y = np.unique(y)
-        if classes is not None:
-            assert set(classes_y) <= set(classes)
-        else:
-            classes = classes_y
-        return len(classes)
+    @property
+    def mode(self):
+        return "classification"
 
     # Only add the 'accuracy' callback if we are not in cropped mode.
     @property

{braindecode-1.3.0.dev169126795 → braindecode-1.3.0.dev169765289}/braindecode/datasets/__init__.py

@@ -11,9 +11,11 @@ from .base import (
 )
 from .bcicomp import BCICompetitionIVDataset4
 from .bids import BIDSDataset, BIDSEpochsDataset
+from .chb_mit import CHBMIT
 from .mne import create_from_mne_epochs, create_from_mne_raw
 from .moabb import BNCI2014_001, HGD, MOABBDataset
 from .nmt import NMT
+from .siena import SIENA
 from .sleep_physio_challe_18 import SleepPhysionetChallenge2018
 from .sleep_physionet import SleepPhysionet
 from .tuh import TUH, TUHAbnormal
@@ -34,7 +36,9 @@ __all__ = [
     "create_from_mne_epochs",
     "TUH",
     "TUHAbnormal",
+    "SIENA",
     "NMT",
+    "CHBMIT",
     "SleepPhysionet",
     "SleepPhysionetChallenge2018",
     "create_from_X_y",

braindecode-1.3.0.dev169765289/braindecode/datasets/chb_mit.py

@@ -0,0 +1,172 @@
+"""
+This dataset is a BIDS-compatible version of the CHB-MIT Scalp EEG Database.
+
+It reorganizes the file structure to comply with the BIDS specification. To this effect:
+
+    The data from subject chb21 was moved to sub-01/ses-02.
+    Metadata was organized according to BIDS.
+    Data in the EEG edf files was modified to keep only the 18 channels from a double banana bipolar montage.
+    Annotations were formatted as BIDS-score compatible `tsv` files.
+"""
+
+# Authors: Dan, Jonathan
+#          Shoeb, Ali (Data collector)
+#          Bruno Aristimunha <b.aristimunha@gmail.com>
+#
+# License: BSD (3-clause)
+from __future__ import annotations
+
+from pathlib import Path
+
+from mne.datasets import fetch_dataset
+
+from braindecode.datasets import BIDSDataset
+from braindecode.datasets.utils import _correct_dataset_path
+
+CHB_MIT_URL = "https://zenodo.org/records/10259996/files/BIDS_CHB-MIT.zip"
+CHB_MIT_archive_name = "chb_mit_bids.zip"
+CHB_MIT_folder_name = "CHB-MIT-BIDS-eeg-dataset"
+CHB_MIT_dataset_name = "CHB-MIT-EEG-Corpus"
+
+CHB_MIT_dataset_params = {
+    "dataset_name": CHB_MIT_dataset_name,
+    "url": CHB_MIT_URL,
+    "archive_name": CHB_MIT_archive_name,
+    "folder_name": CHB_MIT_folder_name,
+    "hash": "078f4e110e40d10fef1a38a892571ad24666c488e8118a01002c9224909256ed",  # sha256
+    "config_key": CHB_MIT_dataset_name,
+}
+
+
+class CHBMIT(BIDSDataset):
+    """The Children's Hospital Boston EEG Dataset.
+
+     This database, collected at the Children's Hospital Boston, consists of EEG recordings
+     from pediatric subjects with intractable seizures. Subjects were monitored for up to
+     several days following withdrawal of anti-seizure medication in order to characterize
+     their seizures and assess their candidacy for surgical intervention.
+
+     Description of the contents of the dataset
+
+     Each folder (sub-01, sub-01, etc.) contains between 9 and 42 continuous .edf
+     files from a single subject. Hardware limitations resulted in gaps between
+     consecutively-numbered .edf files, during which the signals were not recorded;
+     in most cases, the gaps are 10 seconds or less, but occasionally there are much
+     longer gaps. In order to protect the privacy of the subjects, all protected health
+     information (PHI) in the original .edf files has been replaced with surrogate information
+     in the files provided here. Dates in the original .edf files have been replaced by
+     surrogate dates, but the time relationships between the individual files belonging
+     to each case have been preserved. In most cases, the .edf files contain exactly one
+     hour of digitized EEG signals, although those belonging to case sub-10 are two hours
+     long, and those belonging to cases sub-04, sub-06, sub-07, sub-09, and sub-23 are
+     four hours long; occasionally, files in which seizures are recorded are shorter.
+
+     The EEG is recorded at 256 Hz with a 16-bit resolution. The recordings are
+    referenced in a double banana bipolar montage with 18 channels from the 10-20 electrode system.
+
+     This BIDS-compatible version of the dataset was published by Jonathan Dan [Dan2025]_
+     and is based on the original CHB MIT EEG Database [Guttag2010]_, [Shoeb2009]_.
+
+     .. versionadded:: 1.3
+
+     Parameters
+     ----------
+     root : pathlib.Path | str
+         The root of the BIDS path.
+     subjects : str | array-like of str | None
+         The subject ID. Corresponds to "sub".
+     sessions : str | array-like of str | None
+         The acquisition session. Corresponds to "ses".
+     tasks : str | array-like of str | None
+         The experimental task. Corresponds to "task".
+     acquisitions: str | array-like of str | None
+         The acquisition parameters. Corresponds to "acq".
+     runs : str | array-like of str | None
+         The run number. Corresponds to "run".
+     processings : str | array-like of str | None
+         The processing label. Corresponds to "proc".
+     recordings : str | array-like of str | None
+         The recording name. Corresponds to "rec".
+     spaces : str | array-like of str | None
+         The coordinate space for anatomical and sensor location
+         files (e.g., ``*_electrodes.tsv``, ``*_markers.mrk``).
+         Corresponds to "space".
+         Note that valid values for ``space`` must come from a list
+         of BIDS keywords as described in the BIDS specification.
+     splits : str | array-like of str | None
+         The split of the continuous recording file for ``.fif`` data.
+         Corresponds to "split".
+     descriptions : str | array-like of str | None
+         This corresponds to the BIDS entity ``desc``. It is used to provide
+         additional information for derivative data, e.g., preprocessed data
+         may be assigned ``description='cleaned'``.
+     suffixes : str | array-like of str | None
+         The filename suffix. This is the entity after the
+         last ``_`` before the extension. E.g., ``'channels'``.
+         The following filename suffix's are accepted:
+         'meg', 'markers', 'eeg', 'ieeg', 'T1w',
+         'participants', 'scans', 'electrodes', 'coordsystem',
+         'channels', 'events', 'headshape', 'digitizer',
+         'beh', 'physio', 'stim'
+     extensions : str | array-like of str | None
+         The extension of the filename. E.g., ``'.json'``.
+         By default, uses the ones accepted by :func:`mne_bids.read_raw_bids`.
+     datatypes : str | array-like of str | None
+         The BIDS data type, e.g., ``'anat'``, ``'func'``, ``'eeg'``, ``'meg'``,
+         ``'ieeg'``.
+     check : bool
+         If ``True``, only returns paths that conform to BIDS. If ``False``
+         (default), the ``.check`` attribute of the returned
+         :class:`mne_bids.BIDSPath` object will be set to ``True`` for paths that
+         do conform to BIDS, and to ``False`` for those that don't.
+     preload : bool
+         If True, preload the data. Defaults to False.
+     n_jobs : int
+         Number of jobs to run in parallel. Defaults to 1.
+
+     References
+     ----------
+     .. [Guttag2010] Guttag, J. (2010). CHB-MIT Scalp EEG Database (version 1.0.0).
+         PhysioNet. RRID:SCR_007345. https://doi.org/10.13026/C2K01R
+     .. [Shoeb2009] Ali Shoeb. Application of Machine Learning to Epileptic
+         Seizure Onset Detection and Treatment. PhD Thesis,
+         Massachusetts Institute of Technology, September 2009.
+     .. [Dan2025] Dan, J., Pale, U., Amirshahi, A., Cappelletti, W.,
+         Ingolfsson, T. M., Wang, X., ... & Ryvlin, P. (2025).
+         SzCORE: seizure community open-source research evaluatio
+         framework for the validation of electroencephalography-based
+         automated seizure detection algorithms. Epilepsia, 66, 14-24.
+    """
+
+    def __init__(self, root=None, *args, **kwargs):
+        # Download dataset if not present
+        if root is None:
+            path_root = fetch_dataset(
+                dataset_params=CHB_MIT_dataset_params,
+                path=None,
+                processor="unzip",
+                force_update=False,
+            )
+            # First time we fetch the dataset, we need to move the files to the
+            # correct directory.
+            path_root = _correct_dataset_path(
+                path_root, CHB_MIT_archive_name, "BIDS_CHB-MIT"
+            )
+        else:
+            # Validate that the provided root is a valid BIDS dataset
+            if not Path(f"{root}/participants.tsv").exists():
+                raise ValueError(
+                    f"The provided root directory {root} does not contain a valid "
+                    "BIDS dataset (missing participants.tsv). Please ensure the "
+                    "root points directly to the BIDS dataset directory."
+                )
+            path_root = root
+
+        kwargs["root"] = path_root
+
+        super().__init__(
+            *args,
+            extensions=".edf",
+            check=False,
+            **kwargs,
+        )

{braindecode-1.3.0.dev169126795 → braindecode-1.3.0.dev169765289}/braindecode/datasets/hub.py

@@ -21,18 +21,9 @@ import mne
 import numpy as np
 import pandas as pd
 import scipy
+from mne._fiff.meas_info import Info
 from mne.utils import _soft_import
 
-# TODO: Simplify this logic in the future with zarr v3+ only
-# Optional imports for Hub functionality
-try:
-    from numcodecs import Blosc, GZip, Zstd
-
-    NUMCODECS_AVAILABLE = True
-except ImportError:
-    NUMCODECS_AVAILABLE = False
-    Blosc = GZip = Zstd = None
-
 if TYPE_CHECKING:
     from .base import BaseDataset
 
@@ -397,9 +388,8 @@ class HubDatasetMixin:
                 f"{output_path} already exists. Set overwrite=True to replace it."
             )
 
-        # Create zarr store (zarr v2 API)
-        store = zarr.DirectoryStore(str(output_path))
-        root = zarr.group(store=store, overwrite=False)
+        # Create zarr store (zarr v3 API)
+        root = zarr.open(str(output_path), mode="w")
 
         # Validate uniformity across all datasets using shared validation
         dataset_type, _, _ = hub_validation.validate_dataset_uniformity(self.datasets)
@@ -444,7 +434,7 @@ class HubDatasetMixin:
                 data = ds.windows.get_data()
                 metadata = ds.windows.metadata
                 description = ds.description
-                info_dict = _mne_info_to_dict(ds.windows.info)
+                info_dict = ds.windows.info.to_json_dict()
                 target_name = ds.target_name if hasattr(ds, "target_name") else None
 
                 # Save using inlined function
@@ -457,7 +447,7 @@ class HubDatasetMixin:
                 raw = ds.raw
                 metadata = ds.metadata
                 description = ds.description
-                info_dict = _mne_info_to_dict(ds.raw.info)
+                info_dict = ds.raw.info.to_json_dict()
                 targets_from = ds.targets_from
                 last_target_only = ds.last_target_only
 
@@ -477,7 +467,7 @@ class HubDatasetMixin:
                 # Get continuous raw data from RawDataset
                 raw = ds.raw
                 description = ds.description
-                info_dict = _mne_info_to_dict(ds.raw.info)
+                info_dict = ds.raw.info.to_json_dict()
                 target_name = ds.target_name if hasattr(ds, "target_name") else None
 
                 # Save using inlined function
@@ -530,9 +520,8 @@ class HubDatasetMixin:
         if not input_path.exists():
             raise FileNotFoundError(f"{input_path} does not exist.")
 
-        # Open zarr store (zarr v2 API)
-        store = zarr.DirectoryStore(str(input_path))
-        root = zarr.group(store=store)
+        # Open zarr store (zarr v3 API)
+        root = zarr.open(str(input_path), mode="r")
 
         n_datasets = root.attrs["n_datasets"]
         dataset_type = root.attrs["dataset_type"]
@@ -554,7 +543,7 @@ class HubDatasetMixin:
                 )
 
                 # Convert to MNE objects and create dataset
-                info = _dict_to_mne_info(info_dict)
+                info = Info.from_json_dict(info_dict)
                 events = np.column_stack(
                     [
                         metadata["i_start_in_trial"].values,
@@ -580,7 +569,7 @@ class HubDatasetMixin:
 
                 # Convert to MNE objects and create dataset
                 # Data is already in continuous format [n_channels, n_timepoints]
-                info = _dict_to_mne_info(info_dict)
+                info = Info.from_json_dict(info_dict)
                 raw = mne.io.RawArray(data, info)
                 ds = EEGWindowsDataset(
                     raw=raw,
@@ -598,7 +587,7 @@ class HubDatasetMixin:
 
                 # Convert to MNE objects and create dataset
                 # Data is in continuous format [n_channels, n_timepoints]
-                info = _dict_to_mne_info(info_dict)
+                info = Info.from_json_dict(info_dict)
                 raw = mne.io.RawArray(data, info)
                 ds = RawDataset(raw, description)
                 if target_name is not None:
@@ -632,48 +621,22 @@ class HubDatasetMixin:
 # =============================================================================
 
 
-# TODO: remove when this MNE is solved https://github.com/mne-tools/mne-python/issues/13487
-def _mne_info_to_dict(info):
-    """Convert MNE Info object to dictionary for JSON serialization."""
-    return {
-        "ch_names": info["ch_names"],
-        "sfreq": float(info["sfreq"]),
-        "ch_types": [str(ch_type) for ch_type in info.get_channel_types()],
-        "lowpass": float(info["lowpass"]) if info["lowpass"] is not None else None,
-        "highpass": float(info["highpass"]) if info["highpass"] is not None else None,
-    }
-
-
-def _dict_to_mne_info(info_dict):
-    """Convert dictionary back to MNE Info object."""
-    info = mne.create_info(
-        ch_names=info_dict["ch_names"],
-        sfreq=info_dict["sfreq"],
-        ch_types=info_dict["ch_types"],
-    )
-
-    # Use _unlock() to set filter info when reconstructing from saved metadata
-    # This is necessary because MNE protects these fields to prevent users from
-    # setting filter parameters without actually filtering the data
-    with info._unlock():
-        if info_dict.get("lowpass") is not None:
-            info["lowpass"] = info_dict["lowpass"]
-        if info_dict.get("highpass") is not None:
-            info["highpass"] = info_dict["highpass"]
-
-    return info
-
-
 def _save_windows_to_zarr(
     grp, data, metadata, description, info, compressor, target_name
 ):
     """Save windowed data to Zarr group (low-level function)."""
     # Save data with chunking for random access
-    grp.create_dataset(
+    # In Zarr v3, use create_array with compressors parameter
+    data_array = data.astype(np.float32)
+
+    # Zarr v3 expects compressors as a list
+    compressors_list = [compressor] if compressor is not None else None
+
+    grp.create_array(
         "data",
-        data=data.astype(np.float32),
-        chunks=(1, data.shape[1], data.shape[2]),
-        compressor=compressor,
+        data=data_array,
+        chunks=(1, data_array.shape[1], data_array.shape[2]),
+        compressors=compressors_list,
     )
 
     # Save metadata
@@ -704,11 +667,17 @@ def _save_eegwindows_to_zarr(
 
     # Save continuous data with chunking optimized for window extraction
     # Chunk size: all channels, 10000 timepoints for efficient random access
-    grp.create_dataset(
+    # In Zarr v3, use create_array with compressors parameter
+    continuous_float = continuous_data.astype(np.float32)
+
+    # Zarr v3 expects compressors as a list
+    compressors_list = [compressor] if compressor is not None else None
+
+    grp.create_array(
         "data",
-        data=continuous_data.astype(np.float32),
-        chunks=(continuous_data.shape[0], min(10000, continuous_data.shape[1])),
-        compressor=compressor,
+        data=continuous_float,
+        chunks=(continuous_float.shape[0], min(10000, continuous_float.shape[1])),
+        compressors=compressors_list,
     )
 
     # Save metadata
@@ -803,11 +772,17 @@ def _save_raw_to_zarr(grp, raw, description, info, target_name, compressor):
 
     # Save continuous data with chunking optimized for efficient access
     # Chunk size: all channels, 10000 timepoints for efficient random access
-    grp.create_dataset(
+    # In Zarr v3, use create_array with compressors parameter
+    continuous_float = continuous_data.astype(np.float32)
+
+    # Zarr v3 expects compressors as a list
+    compressors_list = [compressor] if compressor is not None else None
+
+    grp.create_array(
         "data",
-        data=continuous_data.astype(np.float32),
-        chunks=(continuous_data.shape[0], min(10000, continuous_data.shape[1])),
-        compressor=compressor,
+        data=continuous_float,
+        chunks=(continuous_float.shape[0], min(10000, continuous_float.shape[1])),
+        compressors=compressors_list,
     )
 
     # Save description
@@ -849,27 +824,35 @@ def _load_raw_from_zarr(grp, preload):
 
 
 def _create_compressor(compression, compression_level):
-    """Create a Zarr compressor object (zarr v2 API)."""
+    """Create a Zarr v3 compressor codec.
+
+    Returns compressor dict compatible with Zarr v3 create_array API.
+
+    Parameters
+    ----------
+    compression : str or None
+        Compression algorithm: "blosc", "zstd", "gzip", or None.
+    compression_level : int
+        Compression level (0-9).
+
+    Returns
+    -------
+    dict or None
+        Compressor configuration dict or None if no compression.
+    """
     if zarr is False:
         raise ImportError(
             "Zarr is not installed. Install with: pip install braindecode[hub]"
         )
 
-    if not NUMCODECS_AVAILABLE:
-        raise ImportError(
-            "numcodecs is not installed. Install with: pip install braindecode[hub]"
-        )
-
-    # Zarr v2 uses numcodecs compressors
-    if compression == "blosc":
-        return Blosc(cname="zstd", clevel=compression_level)
-    elif compression == "zstd":
-        return Zstd(level=compression_level)
-    elif compression == "gzip":
-        return GZip(level=compression_level)
-    else:
+    if compression is None or compression not in ("blosc", "zstd", "gzip"):
         return None
 
+    # Map blosc to zstd (Zarr v3 uses zstd as the primary implementation)
+    name = "zstd" if compression == "blosc" else compression
+
+    return {"name": name, "configuration": {"level": compression_level}}
+
 
 # TODO: improve content
 def _generate_readme_content(