Note
Click here to download the full example code
Plotting functions#
In this tutorial, we will demonstrate how to plot
audio data and spectrograms using the built-in
plotting functions in soundevent.
Audio Arrays#
First lets compute the spectrogram of an audio file.
from soundevent import arrays, audio, data
recording = data.Recording.from_file("sample_audio.wav")
wave = audio.load_recording(recording)
spectrogram = arrays.to_db(
audio.compute_spectrogram(
wave,
window_size=0.064,
hop_size=0.032,
)
)
Both the wave and spectrogram are xarray.DataArray objects. Hence we can use the built-in plotting functions to visualize them.
Out:
And the spectrogram.
Out:
Note
xarray plotting functions are quite flexible and allow you to customize the plot. To see how to use the xarray plotting functions, see the xarray plotting documentation.
Geometries#
To plot geometries, we can use the plot_geometry function in the
soundevent.plot module.
We can plot the different geometries that are defined within soundevent.
time_stamp = data.TimeStamp(coordinates=0.1)
time_interval = data.TimeInterval(coordinates=[0.2, 0.3])
point = data.Point(coordinates=[0.4, 6000])
box = data.BoundingBox(coordinates=[0.5, 3000, 0.6, 5000])
line = data.LineString(coordinates=[[0.7, 2000], [0.8, 3000], [0.9, 8000]])
polygon = data.Polygon(
coordinates=[
[
[1.0, 4000],
[1.1, 5000],
[1.2, 4000],
[1.1, 3000],
[1.0, 4000],
]
]
)
ax = plot.plot_geometry(time_stamp)
plot.plot_geometry(time_interval, ax=ax, color="red")
plot.plot_geometry(point, ax=ax, color="green")
plot.plot_geometry(box, ax=ax, color="blue")
plot.plot_geometry(line, ax=ax, color="purple", linestyle="--")
plot.plot_geometry(polygon, ax=ax, color="orange")
Out:
Sound Event Annotations#
To plot sound event annotations, we can use the plot_annotation function in
the soundevent.plot module.
sound_event_annotations = [
data.SoundEventAnnotation(
tags=[
data.Tag(key="animal", value="dog"),
data.Tag(key="loudness", value="loud"),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[0.58, 100, 0.78, 2000]),
),
),
data.SoundEventAnnotation(
tags=[
data.Tag(key="animal", value="dog"),
data.Tag(key="loudness", value="loud"),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[1.08, 60, 1.34, 1600]),
),
),
data.SoundEventAnnotation(
tags=[
data.Tag(key="animal", value="dog"),
data.Tag(key="loudness", value="medium"),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[1.52, 30, 1.69, 1400]),
),
),
data.SoundEventAnnotation(
tags=[
data.Tag(key="animal", value="dog"),
data.Tag(key="loudness", value="soft"),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[1.98, 30, 2.1, 800]),
),
),
data.SoundEventAnnotation(
tags=[
data.Tag(key="animal", value="dog"),
data.Tag(key="loudness", value="soft"),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[2.45, 30, 2.70, 1300]),
),
),
]
ax = plot.create_axes()
spectrogram.plot(ax=ax, cmap="gray")
plot.plot_annotations(
sound_event_annotations,
ax=ax,
color="red",
add_points=False,
time_offset=0.03,
freq_offset=300,
)
Out:
/opt/hostedtoolcache/Python/3.11.10/x64/lib/python3.11/site-packages/pydantic/main.py:212: DeprecationWarning: The 'key' field is deprecated. Please use 'term' instead.
validated_self = self.__pydantic_validator__.validate_python(data, self_instance=self)
<Axes: title={'center': 'channel = 0'}, xlabel='Time since start of recording\n[s]', ylabel='Frequency [Hz]'>
Sound Event Predictions#
sound_event_predictions = [
data.SoundEventPrediction(
score=0.9,
tags=[
data.PredictedTag(
score=0.95, tag=data.Tag(key="animal", value="dog")
),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[0.58, 100, 0.78, 2000]),
),
),
data.SoundEventPrediction(
score=0.8,
tags=[
data.PredictedTag(
score=0.9, tag=data.Tag(key="animal", value="dog")
),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[1.08, 60, 1.34, 1600]),
),
),
data.SoundEventPrediction(
score=0.4,
tags=[
data.PredictedTag(
score=0.7, tag=data.Tag(key="animal", value="dog")
),
data.PredictedTag(
score=0.3, tag=data.Tag(key="animal", value="cat")
),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[1.52, 30, 1.69, 1400]),
),
),
data.SoundEventPrediction(
score=0.3,
tags=[
data.PredictedTag(
score=0.5, tag=data.Tag(key="animal", value="dog")
),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[1.98, 30, 2.1, 800]),
),
),
data.SoundEventPrediction(
score=0.8,
tags=[
data.PredictedTag(
score=0.4, tag=data.Tag(key="animal", value="dog")
),
],
sound_event=data.SoundEvent(
recording=recording,
geometry=data.BoundingBox(coordinates=[2.45, 30, 2.70, 1300]),
),
),
]
ax = plot.create_axes()
spectrogram.plot(ax=ax, cmap="gray")
plot.plot_predictions(
sound_event_predictions,
ax=ax,
color="red",
add_points=False,
time_offset=0.03,
freq_offset=300,
)
Out:
/opt/hostedtoolcache/Python/3.11.10/x64/lib/python3.11/site-packages/pydantic/main.py:212: DeprecationWarning: The 'key' field is deprecated. Please use 'term' instead.
validated_self = self.__pydantic_validator__.validate_python(data, self_instance=self)
<Axes: title={'center': 'channel = 0'}, xlabel='Time since start of recording\n[s]', ylabel='Frequency [Hz]'>
Total running time of the script: ( 0 minutes 2.463 seconds) Estimated memory usage: 77 MB
Download Python source code: 6_plotting.py