ndx-sound Extension for NWB

NWB extension for sounds.

Installation

pip install ndx-sound

Usage

Python

Add to an NWB file

from pynwb import NWBFile
from scipy.io import wavfile

from ndx_sound import AcousticWaveformSeries

# The file path to the audio file
file_path = "audio_data.wav"

# Read the audio file to get the rate of the recording and the waveform
sampling_rate, samples = wavfile.read(file_path)

# Create an AcousticWaveformSeries object with a given name and description
acoustic_waveform_series = AcousticWaveformSeries(
    name="acoustic_stimulus",
    data=samples,
    rate=sampling_rate,
    description="acoustic stimulus",
)

# Create an NWBFile object where this AcousticWaveformSeries can be added to
nwbfile = NWBFile(
    session_description=...,
    identifier=...,
    session_start_time=...,
)

# If a recording of behavior, add to acquisition
nwbfile.add_acquisition(acoustic_waveform_series)

# If a stimulus, add to stimulus
nwbfile.add_stimulus(acoustic_waveform_series)

Visualization

Static widgets

Use plot_sound to visualize the waveform series and the spectrogram. For longer recordings, specify the time_window argument for the start and end time of the recording to be shown.

from ndx_sound.widgets import plot_sound

plot_sound(nwbfile.stimulus["acoustic_stimulus"])

# Show only from 5 to 15 seconds
plot_sound(nwbfile.stimulus["acoustic_stimulus"], time_window=(5, 15))

Use acoustic_waveform_widget to include an Audio element that plays the sound.

from ndx_sound.widgets import acoustic_waveform_widget

acoustic_waveform_widget(nwbfile.stimulus["acoustic_stimulus"], time_window=(5, 15))

Interactive widgets

Use AcousticWaveformWidget to use a slider for interactively scrolling through the recording and a button for changing the duration of the sound that is being shown.

from ndx_sound.widgets import AcousticWaveformWidget

AcousticWaveformWidget(nwbfile.stimulus["acoustic_stimulus"])

nwbwidgets

Use load_widgets to load the interactive sound widget into nwb2widget.

from ndx_sound.widgets import load_widgets
from nwbwidgets import nwb2widget

load_widgets()

nwb2widget(nwbfile)

nwbwidgets and HDF5IO

When using nwb2widget with an NWB file that is read from disk, make sure to have load_widgets imported within the same Jupyter cell where your data is being loaded.

from pynwb import NWBHDF5IO
from ndx_sound.widgets import load_widgets
from nwbwidgets import nwb2widget

load_widgets()


io = NWBHDF5IO("audio.nwb", mode="r", load_namespaces=True)
nwbfile = io.read()
nwb2widget(nwbfile)

This extension was created using ndx-template.