def scene_metadata_container(cls, filename=None):
from dcase_util.containers import MetaDataContainer
if filename is not None:
return MetaDataContainer().load(filename=filename)
else:
return MetaDataContainer([
{
'filename': cls.audio_filename(),
'scene_label': 'street'
},
{
'filename': 'test1.wav',
'scene_label': 'home'
},
{
'filename': 'test2.wav',
'scene_label': 'office'
},
{
'filename': 'test3.wav',
'scene_label': 'car'
},
{
'filename': 'test4.wav',
'scene_label': 'bus'
},
])
def test_process_events():
meta = MetaDataContainer(content2).process_events(minimum_event_gap=0.5,
minimum_event_length=1.0)
nose.tools.eq_(len(meta), 3)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.5)
nose.tools.eq_(meta[0].offset, 3.0)
nose.tools.eq_(meta[1].filename, 'audio_001.wav')
nose.tools.eq_(meta[1].scene_label, 'office')
nose.tools.eq_(meta[1].event_label, 'speech')
nose.tools.eq_(meta[1].onset, 4.0)
nose.tools.eq_(meta[1].offset, 6.0)
nose.tools.eq_(meta[2].filename, 'audio_001.wav')
nose.tools.eq_(meta[2].scene_label, 'office')
nose.tools.eq_(meta[2].event_label, 'speech')
nose.tools.eq_(meta[2].onset, 7.0)
nose.tools.eq_(meta[2].offset, 8.0)
meta = MetaDataContainer(content2).process_events(minimum_event_gap=1.0,
minimum_event_length=1.0)
nose.tools.eq_(len(meta), 1)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.5)
nose.tools.eq_(meta[0].offset, 8.0)
def test_map_events():
meta = MetaDataContainer(content)
meta_mapped_1 = meta.map_events(target_event_label='activity',
source_event_labels=['speech', 'printer'])
nose.tools.eq_(len(meta_mapped_1), 4)
meta_mapped_2 = meta.map_events(target_event_label='activity')
nose.tools.eq_(len(meta_mapped_2), 5)
def test_process_events():
meta = MetaDataContainer(content2).process_events(minimum_event_gap=0.5,
minimum_event_length=1.0)
nose.tools.eq_(len(meta), 3)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.5)
nose.tools.eq_(meta[0].offset, 3.0)
nose.tools.eq_(meta[1].filename, 'audio_001.wav')
nose.tools.eq_(meta[1].scene_label, 'office')
nose.tools.eq_(meta[1].event_label, 'speech')
nose.tools.eq_(meta[1].onset, 4.0)
nose.tools.eq_(meta[1].offset, 6.0)
nose.tools.eq_(meta[2].filename, 'audio_001.wav')
nose.tools.eq_(meta[2].scene_label, 'office')
nose.tools.eq_(meta[2].event_label, 'speech')
nose.tools.eq_(meta[2].onset, 7.0)
nose.tools.eq_(meta[2].offset, 8.0)
meta = MetaDataContainer(content2).process_events(minimum_event_gap=1.0,
minimum_event_length=1.0)
nose.tools.eq_(len(meta), 1)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.5)
nose.tools.eq_(meta[0].offset, 8.0)
meta = MetaDataContainer([
{
'filename': 'audio_001.wav',
'scene_label': 'office',
'event_label': 'mouse clicking',
'onset': 3.0,
'offset': 5.0,
'identifier': 'A001'
},
{
'filename': 'audio_001.wav',
'scene_label': 'office',
'event_label': 'mouse clicking',
'onset': 13.0,
'offset': 13.0,
'identifier': 'A001'
},
]).process_events(minimum_event_gap=1.0,
minimum_event_length=numpy.spacing(1))
nose.tools.eq_(len(meta), 1)
def __init__(self, *args, **kwargs):
"""Constructor"""
# Run ProcessorMixin init
ProcessorMixin.__init__(self, *args, **kwargs)
# Run MetaDataContainer init
MetaDataContainer.__init__(self, **kwargs)
# Run super init to call init of mixins too
super(MetadataReadingProcessor, self).__init__(*args, **kwargs)
def test_addition():
meta = MetaDataContainer(content)
meta2 = MetaDataContainer(content2)
meta += meta2
nose.tools.eq_(len(meta), 9)
nose.tools.eq_(meta[8].filename, 'audio_001.wav')
nose.tools.eq_(meta[8].scene_label, 'office')
nose.tools.eq_(meta[8].event_label, 'speech')
nose.tools.eq_(meta[8].onset, 7.0)
nose.tools.eq_(meta[8].offset, 8.0)
def tag_metadata_container(cls, filename=None):
from dcase_util.containers import MetaDataContainer
if filename is not None:
return MetaDataContainer().load(filename=filename)
else:
return MetaDataContainer([{
'filename': cls.audio_filename(),
'scene_label': 'street',
'tags': 'car, bicycle',
}])
def test_intersection():
data1 = MetaDataContainer(content)
data2 = MetaDataContainer([{
'filename': 'audio_001.wav',
'scene_label': 'office',
'event_label': 'speech',
'onset': 1.0,
'offset': 10.0,
'identifier': 'A001'
}])
intersection = data1.intersection(data2)
nose.tools.eq_(len(intersection), 1)
nose.tools.eq_(intersection[0].filename, 'audio_001.wav')
def event_metadata_container(cls, filename=None):
from dcase_util.containers import MetaDataContainer
if filename is not None:
return MetaDataContainer().load(
filename=filename
)
else:
return MetaDataContainer(
[
{
'filename': cls.audio_filename(),
'scene_label': 'street',
'event_label': 'car',
'onset': 0.0,
'offset': 10.0
},
{
'filename': cls.audio_filename(),
'scene_label': 'street',
'event_label': 'bicycle',
'onset': 0.0,
'offset': 3.641
},
{
'filename': 'test1.wav',
'scene_label': 'street',
'event_label': 'car',
'onset': 3.0,
'offset': 5.0
},
{
'filename': 'test1.wav',
'scene_label': 'street',
'event_label': 'bicycle',
'onset': 6.0,
'offset': 7.0
},
{
'filename': 'test1.wav',
'scene_label': 'street',
'event_label': 'car',
'onset': 7.0,
'offset': 8.0
},
]
)
def test_pad():
minimal_event_list = [{
'event_label': 'A',
'onset': 0,
'offset': 1,
}, {
'event_label': 'A',
'onset': 5,
'offset': 15,
}, {
'event_label': 'B',
'onset': 1,
'offset': 2,
}, {
'event_label': 'B',
'onset': 4,
'offset': 5,
}, {
'event_label': 'C',
'onset': 7,
'offset': 12,
}]
meta = MetaDataContainer(minimal_event_list)
target_event_roll = numpy.array([
[1., 0., 0.], # 0
[0., 1., 0.], # 1
[0., 0., 0.], # 2
[0., 0., 0.], # 3
[0., 1., 0.], # 4
[1., 0., 0.], # 5
[1., 0., 0.], # 6
[1., 0., 1.], # 7
[1., 0., 1.], # 8
[1., 0., 1.], # 9
[1., 0., 1.], # 10
[1., 0., 1.], # 11
[1., 0., 0.], # 12
[1., 0., 0.], # 13
[1., 0., 0.], # 14
[0., 0., 0.], # 15
]).T
# Test #1
roller = EventRollEncoder(label_list=['A', 'B', 'C'], time_resolution=1.0)
event_roll = roller.encode(metadata_container=meta, length_seconds=16)
numpy.testing.assert_array_equal(target_event_roll, event_roll.data)
nose.tools.assert_equal(event_roll.shape[0], target_event_roll.shape[0])
nose.tools.assert_equal(event_roll.shape[1], target_event_roll.shape[1])
padded_event_roll = roller.pad(length=18)
nose.tools.assert_equal(padded_event_roll.length, 18)
nose.tools.assert_equal(padded_event_roll.shape[1], event_roll.shape[1])
padded_event_roll = roller.pad(length=10)
nose.tools.assert_equal(padded_event_roll.length, 10)
nose.tools.assert_equal(padded_event_roll.shape[1], event_roll.shape[1])
def test_event_inactivity():
meta = MetaDataContainer(content)
meta_inactivity = meta.event_inactivity()
nose.tools.eq_(len(meta_inactivity), 3)
meta = MetaDataContainer([
{
'filename': 'audio_001.wav',
'scene_label': 'office',
'event_label': 'mouse clicking',
'onset': 3.0,
'offset': 5.0,
'identifier': 'A001'
},
{
'filename': 'audio_001.wav',
'scene_label': 'office',
'event_label': 'mouse clicking',
'onset': 13.0,
'offset': 13.0,
'identifier': 'A001'
},
])
meta_inactivity = meta.event_inactivity(
duration_list={'audio_001.wav': 20.0})
nose.tools.eq_(meta_inactivity[0].onset, 0.00)
nose.tools.eq_(meta_inactivity[0].offset, 3.00)
nose.tools.eq_(meta_inactivity[1].onset, 5.00)
nose.tools.eq_(meta_inactivity[1].offset, 20.00)
def test_construction():
minimal_event_list = [{
'scene_label': 'A'
}, {
'scene_label': 'A'
}, {
'scene_label': 'B'
}, {
'scene_label': 'B'
}, {
'scene_label': 'C'
}]
meta = MetaDataContainer(minimal_event_list)
target_binary_matrix = numpy.array([
[1., 0., 0.], # 0
[1., 0., 0.], # 1
[1., 0., 0.], # 2
]).T
# Test #1
binary_matrix = OneHotEncoder(label_list=['A', 'B', 'C'],
time_resolution=1.0).encode(
label='A',
length_seconds=3,
)
numpy.testing.assert_array_equal(target_binary_matrix, binary_matrix.data)
nose.tools.assert_equal(binary_matrix.shape[0],
target_binary_matrix.shape[0])
nose.tools.assert_equal(binary_matrix.shape[1],
target_binary_matrix.shape[1])
target_binary_matrix = numpy.array([
[0., 1., 0.], # 0
[0., 1., 0.], # 1
[0., 1., 0.], # 2
]).T
# Test #1
binary_matrix = OneHotEncoder(label_list=['A', 'B', 'C'],
time_resolution=1.0).encode(
label='B',
length_seconds=3,
)
numpy.testing.assert_array_equal(target_binary_matrix, binary_matrix.data)
nose.tools.assert_equal(binary_matrix.shape[0],
target_binary_matrix.shape[0])
nose.tools.assert_equal(binary_matrix.shape[1],
target_binary_matrix.shape[1])
def test_construction():
minimal_event_list = [{
'event_label': 'A',
'onset': 0,
'offset': 1,
}, {
'event_label': 'A',
'onset': 5,
'offset': 15,
}, {
'event_label': 'B',
'onset': 1,
'offset': 2,
}, {
'event_label': 'B',
'onset': 4,
'offset': 5,
}, {
'event_label': 'C',
'onset': 7,
'offset': 12,
}]
meta = MetaDataContainer(minimal_event_list)
target_event_roll = numpy.array([
[1., 0., 0.], # 0
[0., 1., 0.], # 1
[0., 0., 0.], # 2
[0., 0., 0.], # 3
[0., 1., 0.], # 4
[1., 0., 0.], # 5
[1., 0., 0.], # 6
[1., 0., 1.], # 7
[1., 0., 1.], # 8
[1., 0., 1.], # 9
[1., 0., 1.], # 10
[1., 0., 1.], # 11
[1., 0., 0.], # 12
[1., 0., 0.], # 13
[1., 0., 0.], # 14
]).T
# Test #1
event_roll = EventRollEncoder(
label_list=['A', 'B', 'C'], time_resolution=1.0,
label='event_label').encode(metadata_container=meta)
numpy.testing.assert_array_equal(target_event_roll, event_roll.data)
nose.tools.assert_equal(event_roll.shape[0], target_event_roll.shape[0])
nose.tools.assert_equal(event_roll.shape[1], target_event_roll.shape[1])
def __init__(self, **kwargs):
"""Constructor
Parameters
----------
time_resolution : float
Hop length in seconds
"""
super(Masker, self).__init__(**kwargs)
# Initialize mask events
from dcase_util.containers import MetaDataContainer
self.mask_events = MetaDataContainer()
def test_content():
meta = MetaDataContainer(content)
nose.tools.eq_(len(meta), 5)
# Test content
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.0)
nose.tools.eq_(meta[0].offset, 10.0)
nose.tools.eq_(meta[4].filename, 'audio_002.wav')
nose.tools.eq_(meta[4].scene_label, 'meeting')
nose.tools.eq_(meta[4].event_label, 'printer')
nose.tools.eq_(meta[4].onset, 5.0)
nose.tools.eq_(meta[4].offset, 7.0)
def test_dataset():
d = Dataset()
d.meta_container = MetaDataContainer(content)
nose.tools.eq_(d.scene_labels(), ['meeting', 'office'])
nose.tools.eq_(d.scene_label_count(), 2)
nose.tools.eq_(d.event_labels(), ['mouse clicking', 'printer', 'speech'])
nose.tools.eq_(d.event_label_count(), 3)
nose.tools.eq_(d.meta_count, len(content))
nose.tools.eq_(d.meta, content)
nose.tools.eq_(d.file_meta(filename='audio_001.wav'),
[content[0], content[1], content[2]])
nose.tools.eq_(d.file_meta(filename='audio_002.wav'),
[content[3], content[4]])
def test_add_time_offset():
meta = MetaDataContainer(content2).add_time(time=2.0)
nose.tools.eq_(len(meta), 4)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 3.0)
nose.tools.eq_(meta[0].offset, 3.2)
nose.tools.eq_(meta[3].filename, 'audio_001.wav')
nose.tools.eq_(meta[3].scene_label, 'office')
nose.tools.eq_(meta[3].event_label, 'speech')
nose.tools.eq_(meta[3].onset, 9.0)
nose.tools.eq_(meta[3].offset, 10.0)
def fill_event_gaps(events, duration):
filled_events = MetaDataContainer()
if len(events) == 0:
filled_events.append(
MetaDataItem({
'onset': 0.0,
'offset': duration,
'event_label': '-'
}))
return filled_events
sorted_events = sorted(events)
last_offset = 0.0
unannotated_segments = [[0.0, duration]]
for i, e in enumerate(sorted_events):
for i, seg in enumerate(unannotated_segments):
if e.onset < seg[1] and e.offset > seg[0]:
if seg[0] >= e.onset and seg[1] <= e.offset:
unannotated_segments.pop(i)
elif seg[0] >= e.onset:
unannotated_segments[i][0] = e.offset
elif seg[1] <= e.offset:
unannotated_segments[i][1] = e.onset
else:
unannotated_segments.append([e.offset, seg[1]])
unannotated_segments[i][1] = e.onset
filled_events.append(
MetaDataItem({
'onset': e.onset,
'offset': e.offset,
'event_label': e.event_label
}))
for seg in unannotated_segments:
filled_events.append(
MetaDataItem({
'onset': seg[0],
'offset': seg[1],
'event_label': '-'
}))
return filled_events
def system_output_file(self, filename):
"""Checks that MetaDataContainer file exists, and can be parsed.
Parameters
----------
filename : str
File path
Raises
------
ImportError:
Error if file format specific module cannot be imported
IOError:
File does not exists or has unknown file format
Returns
-------
MetaDataContainer or bool
"""
if not os.path.isfile(filename):
self.error_log.append(u'File missing [{filename:s}]'.format(
filename=filename
))
return False
else:
try:
if self.task == 'TAG':
data = ProbabilityContainer().load(filename=filename)
else:
data = MetaDataContainer().load(filename=filename)
return data
except Exception:
self.error_log.append(u'File cannot be parsed [{filename:s}]'.format(
filename=filename
))
def test_max_event_offset():
nose.tools.eq_(MetaDataContainer(content).max_offset, 10)
def test_unique_files():
files = MetaDataContainer(content).unique_files
nose.tools.eq_(len(files), 2)
nose.tools.eq_(files[0], 'audio_001.wav')
nose.tools.eq_(files[1], 'audio_002.wav')
def test_unique_event_labels():
events = MetaDataContainer(content).unique_event_labels
nose.tools.eq_(len(events), 3)
nose.tools.eq_(events[0], 'mouse clicking')
nose.tools.eq_(events[1], 'printer')
nose.tools.eq_(events[2], 'speech')
def test_unique_scene_labels():
scenes = MetaDataContainer(content).unique_scene_labels
nose.tools.eq_(len(scenes), 2)
nose.tools.eq_(scenes[0], 'meeting')
nose.tools.eq_(scenes[1], 'office')
def test_event_label_count():
nose.tools.eq_(MetaDataContainer(content).event_label_count, 3)
def test_scene_label_count():
nose.tools.eq_(MetaDataContainer(content).scene_label_count, 2)
def test_event_count():
nose.tools.eq_(MetaDataContainer(content).event_count, len(content))
def test_formats():
delimiters = [',', ';', '\t']
for delimiter in delimiters:
tmp = tempfile.NamedTemporaryFile('r+',
suffix='.txt',
dir='/tmp',
delete=False)
try:
tmp.write('0.5' + delimiter + '0.7\n')
tmp.write('2.5' + delimiter + '2.7\n')
tmp.close()
item = MetaDataContainer().load(filename=tmp.name)[0]
nose.tools.eq_(item.onset, 0.5)
nose.tools.eq_(item.offset, 0.7)
finally:
os.unlink(tmp.name)
tmp = tempfile.NamedTemporaryFile('r+',
suffix='.txt',
dir='/tmp',
delete=False)
try:
tmp.write('0.5' + delimiter + '0.7' + delimiter + 'event\n')
tmp.write('2.5' + delimiter + '2.7' + delimiter + 'event\n')
tmp.close()
item = MetaDataContainer().load(filename=tmp.name)[0]
nose.tools.eq_(item.onset, 0.5)
nose.tools.eq_(item.offset, 0.7)
nose.tools.eq_(item.event_label, 'event')
finally:
os.unlink(tmp.name)
tmp = tempfile.NamedTemporaryFile('r+',
suffix='.txt',
dir='/tmp',
delete=False)
try:
tmp.write('file.wav' + delimiter + 'scene' + delimiter + '0.5' +
delimiter + '0.7' + delimiter + 'event\n')
tmp.write('file.wav' + delimiter + 'scene' + delimiter + '0.5' +
delimiter + '0.7' + delimiter + 'event\n')
tmp.close()
item = MetaDataContainer().load(filename=tmp.name)[0]
nose.tools.eq_(item.onset, 0.5)
nose.tools.eq_(item.offset, 0.7)
nose.tools.eq_(item.event_label, 'event')
nose.tools.eq_(item.filename, 'file.wav')
nose.tools.eq_(item.scene_label, 'scene')
finally:
os.unlink(tmp.name)
tmp = tempfile.NamedTemporaryFile('r+',
suffix='.txt',
dir='/tmp',
delete=False)
try:
tmp.write('file.wav' + delimiter + 'scene' + delimiter + '0.5' +
delimiter + '0.7' + delimiter + 'event' + delimiter +
'm' + delimiter + 'a1\n')
tmp.write('file.wav' + delimiter + 'scene' + delimiter + '0.5' +
delimiter + '0.7' + delimiter + 'event' + delimiter +
'm' + delimiter + 'a2\n')
tmp.close()
item = MetaDataContainer().load(filename=tmp.name)[0]
nose.tools.eq_(item.onset, 0.5)
nose.tools.eq_(item.offset, 0.7)
nose.tools.eq_(item.event_label, 'event')
nose.tools.eq_(item.filename, 'file.wav')
nose.tools.eq_(item.scene_label, 'scene')
nose.tools.eq_(item.identifier, 'a1')
nose.tools.eq_(item.source_label, 'm')
finally:
os.unlink(tmp.name)
def test_filter_time_segment():
# Case 1
meta = MetaDataContainer(content).filter_time_segment(
filename='audio_001.wav',
start=1.0,
stop=3.5,
zero_time=True,
trim=True,
)
nose.tools.eq_(len(meta), 2)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 0.0)
nose.tools.eq_(meta[0].offset, 2.5)
nose.tools.eq_(meta[1].filename, 'audio_001.wav')
nose.tools.eq_(meta[1].scene_label, 'office')
nose.tools.eq_(meta[1].event_label, 'mouse clicking')
nose.tools.eq_(meta[1].onset, 2.0)
nose.tools.eq_(meta[1].offset, 2.5)
# Case 2
meta = MetaDataContainer(content).filter_time_segment(
filename='audio_001.wav',
start=1.0,
stop=3.5,
zero_time=False,
trim=True,
)
nose.tools.eq_(len(meta), 2)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.0)
nose.tools.eq_(meta[0].offset, 3.5)
nose.tools.eq_(meta[1].filename, 'audio_001.wav')
nose.tools.eq_(meta[1].scene_label, 'office')
nose.tools.eq_(meta[1].event_label, 'mouse clicking')
nose.tools.eq_(meta[1].onset, 3.0)
nose.tools.eq_(meta[1].offset, 3.5)
# Case 3
meta = MetaDataContainer(content).filter_time_segment(
filename='audio_001.wav',
start=1.0,
stop=3.5,
zero_time=False,
trim=False,
)
nose.tools.eq_(len(meta), 2)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.0)
nose.tools.eq_(meta[0].offset, 10.0)
nose.tools.eq_(meta[1].filename, 'audio_001.wav')
nose.tools.eq_(meta[1].scene_label, 'office')
nose.tools.eq_(meta[1].event_label, 'mouse clicking')
nose.tools.eq_(meta[1].onset, 3.0)
nose.tools.eq_(meta[1].offset, 5.0)
def test_filter():
# Test filter by file
meta = MetaDataContainer(content).filter(filename='audio_002.wav')
nose.tools.eq_(len(meta), 2)
nose.tools.eq_(meta[0].filename, 'audio_002.wav')
nose.tools.eq_(meta[0].scene_label, 'meeting')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.0)
nose.tools.eq_(meta[0].offset, 9.0)
nose.tools.eq_(meta[1].filename, 'audio_002.wav')
nose.tools.eq_(meta[1].scene_label, 'meeting')
nose.tools.eq_(meta[1].event_label, 'printer')
nose.tools.eq_(meta[1].onset, 5.0)
nose.tools.eq_(meta[1].offset, 7.0)
# Test filter by scene_label
meta = MetaDataContainer(content).filter(scene_label='office')
nose.tools.eq_(len(meta), 3)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.0)
nose.tools.eq_(meta[0].offset, 10.0)
nose.tools.eq_(meta[1].filename, 'audio_001.wav')
nose.tools.eq_(meta[1].scene_label, 'office')
nose.tools.eq_(meta[1].event_label, 'mouse clicking')
nose.tools.eq_(meta[1].onset, 3.0)
nose.tools.eq_(meta[1].offset, 5.0)
meta = MetaDataContainer(content).filter(scene_list=['meeting'])
nose.tools.eq_(len(meta), 2)
# Test filter by event_label
meta = MetaDataContainer(content).filter(event_label='speech')
nose.tools.eq_(len(meta), 2)
nose.tools.eq_(meta[0].filename, 'audio_001.wav')
nose.tools.eq_(meta[0].scene_label, 'office')
nose.tools.eq_(meta[0].event_label, 'speech')
nose.tools.eq_(meta[0].onset, 1.0)
nose.tools.eq_(meta[0].offset, 10.0)
nose.tools.eq_(meta[1].filename, 'audio_002.wav')
nose.tools.eq_(meta[1].scene_label, 'meeting')
nose.tools.eq_(meta[1].event_label, 'speech')
nose.tools.eq_(meta[1].onset, 1.0)
nose.tools.eq_(meta[1].offset, 9.0)
meta = MetaDataContainer(content).filter(event_list=['speech', 'printer'])
nose.tools.eq_(len(meta), 4)
# Test filter by tags
meta = MetaDataContainer(content3).filter(tag='tag1')
nose.tools.eq_(len(meta), 2)
meta = MetaDataContainer(content3).filter(tag_list=['tag1', 'tag3'])
nose.tools.eq_(len(meta), 3)
def prepare(self):
"""Prepare dataset for the usage.
Returns
-------
self
"""
if not self.meta_container.exists():
scene_label = 'home'
evaluation_chunks = ListDictContainer(
filename=os.path.join(self.local_path, 'chime_home',
'evaluation_chunks_refined.csv')).load(
fields=['id', 'filename', 'set_id'])
audio_files = {}
for item in dcase_cross_val_data:
audio_filename = os.path.join(
'chime_home', 'chunks',
item['filename'] + self.sample_mode + '.wav')
annotation_filename = os.path.join('chime_home', 'chunks',
item['filename'] + '.csv')
if audio_filename not in audio_files:
audio_files[audio_filename] = {
'audio': audio_filename,
'meta': annotation_filename
}
meta_data = MetaDataContainer()
for audio_filename, data in iteritems(audio_files):
current_meta_data = DictContainer(filename=os.path.join(
self.local_path, data['meta'])).load()
tags = []
for i, tag in enumerate(current_meta_data['majorityvote']):
if tag != 'S' and tag != 'U':
tags.append(self.tagcode_to_taglabel(tag))
name = os.path.split(audio_filename)[1]
segment_name = name[0:name.find('_chunk')]
chunk_name = name[name.find('_chunk') + 1:].split('.')[0]
item = MetaDataItem({
'filename': audio_filename,
'scene_label': scene_label,
'tags': ';'.join(tags) + ';',
'identifier': segment_name
})
self.process_meta_item(item=item, absolute_path=False)
meta_data.append(item)
# Save meta
meta_data.save(filename=self.meta_file)
# Load meta and cross validation
self.load()
all_folds_found = True
train_filename = self.evaluation_setup_filename(setup_part='train')
test_filename = self.evaluation_setup_filename(setup_part='test')
eval_filename = self.evaluation_setup_filename(setup_part='evaluate')
if not os.path.isfile(train_filename):
all_folds_found = False
if not os.path.isfile(test_filename):
all_folds_found = False
if not os.path.isfile(eval_filename):
all_folds_found = False
if not all_folds_found:
Path().makedirs(path=self.evaluation_setup_path)
# Train
train_filename = self.evaluation_setup_filename(setup_part='train')
train_meta = MetaDataContainer(filename=train_filename)
for filename in self.train_files():
train_meta.append(self.file_meta(filename)[0])
train_meta.save()
# Test
test_filename = self.evaluation_setup_filename(setup_part='test')
test_meta = MetaDataContainer(filename=test_filename)
for filename in self.test_files():
test_meta.append(
MetaDataItem(
{'filename':
self.absolute_to_relative_path(filename)}))
test_meta.save()
# Evaluate
eval_filename = self.evaluation_setup_filename(
setup_part='evaluate')
eval_meta = MetaDataContainer(filename=eval_filename)
for filename in self.test_files():
eval_meta.append(self.file_meta(filename)[0])
eval_meta.save()
# Load meta and cross validation
self.load()
return self
