← Back to API reference

neuralset.events.etypes.MneRaw

pydantic model neuralset.events.etypes.MneRaw

Brain recording saved as MNE Raw object.

Base class for neurophysiological recordings (MEG, EEG, etc.) that can be loaded using MNE-Python. Supports chunking via _split() inherited from BaseSplittableEvent.

Parameters:

subject (str) – Subject identifier (required, cannot be empty)

Notes

• Automatically detects frequency and duration from file metadata

• If start is "auto", auto-resolves from raw.first_samp / sfreq (use for FIF files with nonzero first_samp)

• Guards against start=0 when raw.first_samp > 0

• Subject ID is cast to string to handle numeric IDs from dataframes

• Like Audio/Video, read() crops the raw to [offset, offset+duration] so that only the chunk’s data is loaded into memory.

Examples

meg = Meg(start="auto", timeline="scan1",
         filepath="data_raw.fif", subject="sub-01")
raw = meg.read()  # Returns mne.io.Raw object

Fields:

• subject (Annotated[str, pydantic.functional_validators.BeforeValidator(func=neuralset.base._int_cast, json_schema_input_type=PydanticUndefined)])

field subject: Annotated[str, BeforeValidator(func=_int_cast, json_schema_input_type=PydanticUndefined)] = ''

read() → Any

Read and return the data from the file or method URI.

Returns:

The loaded data (type depends on the specific Event subclass)

Return type:

Any

Examples

audio_event = Audio(start=0, timeline="t1", filepath="audio.wav")
audio_tensor = audio_event.read()  # Returns torch.Tensor

requirements: tp.ClassVar[tuple[str, ...]] = ()

type: tp.ClassVar[str] = 'MneRaw'

← Back to API reference