def test_short_background_fg_events():
foreground_fd = os.path.join(absolute_dir_path, "material", "soundbank",
"foreground")
background_fd = os.path.join(absolute_dir_path, "material", "soundbank",
"background")
sc = Scaper(10, foreground_fd, background_fd, random_state=2020)
sc.sr = 16000
sc.add_background(
("const", "label"),
("const",
os.path.join(background_fd, "label", "noise-free-sound-0055.wav")),
("const", 0))
fname = "test_bg_fg"
fpath = os.path.join(absolute_dir_path, "generated", "short_bg_scaper",
fname)
sc.add_event(
("const", "label"),
("const", os.path.join(foreground_fd, "label", "26104_0.wav")),
("const", 0), ("const", 5), ("const", 5), ("const", 6), ("const", 0),
("const", 1))
sc.generate(f"{fpath}.wav", f"{fpath}.jams")
audio_g, sr_g = soundfile.read(f"{fpath}.wav")
audio_s, sr_s = soundfile.read(
os.path.join(absolute_dir_path, "material", "scaper", f"{fname}.wav"))
print(f"audio gen: {audio_g}")
print(f"audio source: {audio_s}")
print(f"shapes, source: {audio_s.shape}, gen: {audio_g.shape}")
assert audio_g.shape == audio_s.shape
def __test_gini_from_event_times(event_time_list,
expected_gini,
hop_size=0.01):
print(event_time_list)
# create scaper
sc = Scaper(10.0, FG_PATH, BG_PATH)
# add background
sc.add_background(label=('choose', []),
source_file=('choose', []),
source_time=('const', 0))
# add foreground events based on the event time list
# always use siren file since it is 26 s long, so we can choose the
# event duration flexibly
for onset, offset in event_time_list:
sc.add_event(label=('const', 'siren'),
source_file=('const', SIREN_FILE),
source_time=('const', 0),
event_time=('const', onset),
event_duration=('const', offset - onset),
snr=('uniform', 6, 30),
pitch_shift=('uniform', -3, 3),
time_stretch=None)
jam = sc._instantiate()
ann = jam.annotations[0]
gini = polyphony_gini(ann, hop_size=hop_size)
print(gini, expected_gini)
assert np.allclose([gini], [expected_gini], atol=1e-5)
def __test_gini_from_event_times(event_time_list, expected_gini,
hop_size=0.01):
print(event_time_list)
# create scaper
sc = Scaper(10.0, FG_PATH, BG_PATH)
# add background
sc.add_background(label=('choose', []),
source_file=('choose', []),
source_time=('const', 0))
# add foreground events based on the event time list
# always use siren file since it is 26 s long, so we can choose the
# event duration flexibly
for onset, offset in event_time_list:
sc.add_event(label=('const', 'siren'),
source_file=('const', SIREN_FILE),
source_time=('const', 0),
event_time=('const', onset),
event_duration=('const', offset - onset),
snr=('uniform', 6, 30),
pitch_shift=('uniform', -3, 3),
time_stretch=None)
jam = sc._instantiate()
ann = jam.annotations[0]
gini = polyphony_gini(ann, hop_size=hop_size)
print(gini, expected_gini)
assert np.allclose([gini], [expected_gini], atol=1e-5)
def mix_with_scaper(num_mixtures,
foreground_path,
background_path,
scene_duration,
sample_rate,
target_folder,
event_parameters,
num_sources=None,
labels=None,
coherent=False,
allow_repeated_label=False,
ref_db=-40,
bitdepth=16,
seed=0,
num_workers=1):
nussl.utils.seed(seed)
os.makedirs(target_folder, exist_ok=True)
scaper_seed = np.random.randint(100)
logging.info('Starting mixing.')
if num_sources is None and labels is None:
raise ValueError("One of labels or num_sources must be set!")
if coherent and labels is None:
raise ValueError("Coherent mixing requires explicit labels!")
generators = []
if background_path is None:
background_path = foreground_path
for i in range(num_mixtures):
sc = Scaper(
scene_duration,
fg_path=foreground_path,
bg_path=background_path,
random_state=scaper_seed,
)
sc.ref_db = ref_db
sc.sr = sample_rate
sc.bitdepth = bitdepth
generators.append(sc)
scaper_seed += 1
mix_func = make_one_mixture_coherent if coherent else make_one_mixture
def arg_tuple(i):
_args = (generators[i], os.path.join(target_folder, f'{i:08d}.wav'),
labels if coherent else num_sources, event_parameters,
allow_repeated_label)
return _args
args = [arg_tuple(i) for i in range(num_mixtures)]
# do one by itself for testing
mix_func(*args[0])
args = list(zip(*args[1:]))
args = [list(a) for a in args]
# now do the rest in parallel
p_tqdm.p_map(mix_func, *args, num_cpus=num_workers)
def test_polyphony_gini():
'''
Test computation of polyphony gini
'''
# Annotation must have namespace scaper, otherwise raise error
ann = jams.Annotation('tag_open', duration=10)
gini = pytest.raises(ScaperError, polyphony_gini, ann)
# Annotation without duration set should raise error
ann = jams.Annotation('scaper', duration=None)
gini = pytest.raises(ScaperError, polyphony_gini, ann)
# Annotation with no foreground events returns a gini of 0
sc = Scaper(10.0, FG_PATH, BG_PATH)
# add background
sc.add_background(label=('choose', []),
source_file=('choose', []),
source_time=('const', 0))
jam = sc._instantiate()
ann = jam.annotations[0]
gini = polyphony_gini(ann)
assert gini == 0
def __test_gini_from_event_times(event_time_list,
expected_gini,
hop_size=0.01):
print(event_time_list)
# create scaper
sc = Scaper(10.0, FG_PATH, BG_PATH)
# add background
sc.add_background(label=('choose', []),
source_file=('choose', []),
source_time=('const', 0))
# add foreground events based on the event time list
# always use siren file since it is 26 s long, so we can choose the
# event duration flexibly
for onset, offset in event_time_list:
sc.add_event(label=('const', 'siren'),
source_file=('const', SIREN_FILE),
source_time=('const', 0),
event_time=('const', onset),
event_duration=('const', offset - onset),
snr=('uniform', 6, 30),
pitch_shift=('uniform', -3, 3),
time_stretch=None)
jam = sc._instantiate()
ann = jam.annotations[0]
gini = polyphony_gini(ann, hop_size=hop_size)
print(gini, expected_gini)
assert np.allclose([gini], [expected_gini], atol=1e-5)
event_time_lists = ([[], [(0, 1)], [(0, 5), (5, 10)],
[(0, 10), (3, 7), (4, 6)]])
expected_ginis = [0, 0.1, 1, 0.75]
for etl, g in zip(event_time_lists, expected_ginis):
__test_gini_from_event_times(etl, g, hop_size=0.01)
for etl, g in zip(event_time_lists, expected_ginis):
__test_gini_from_event_times(etl, g, hop_size=1.0)
def scaper_mix(
mixture_parameters,
sample_rate,
event_parameters=None,
coherent=False,
ref_db=-40,
bitdepth=16,
seed=0,
num_workers=1,
):
np.random.seed(seed)
logging.info('Making JAMS files')
for key, params in mixture_parameters.items():
if 'event_parameters' in params:
_event_parameters = params['event_parameters']
else:
_event_parameters = event_parameters
generators = []
logging.info('Making generators')
bg_path = params['background_path']
if not bg_path:
bg_path = params['foreground_path']
for i in tqdm(range(params['num_mixtures'])):
sc = Scaper(
params['scene_duration'],
fg_path=params['foreground_path'],
bg_path=bg_path,
random_state=np.random.randint(params['num_mixtures']*10)
)
sc.ref_db = ref_db
sc.sr = sample_rate
sc.bitdepth = bitdepth
generators.append(sc)
os.makedirs(params['target_path'], exist_ok=True)
timings = {}
silent_files = {'default': True}
if coherent:
args = [
{
'path_to_file': os.path.join(params['target_path'], f'{i:08d}.wav'),
'sc': generators[i],
'labels': params['labels'],
'event_parameters': _event_parameters,
} for i in range(params['num_mixtures'])
]
mix_func = make_one_mixture_coherent
else:
args = [
{
'path_to_file': os.path.join(params['target_path'], f'{i:08d}.wav'),
'sc': generators[i],
'num_sources': params['num_sources'],
'event_parameters': _event_parameters,
} for i in range(params['num_mixtures'])
]
mix_func = make_one_mixture
parallel_process(args, mix_func, n_jobs=num_workers, front_num=1, use_kwargs=True)
logging.info(f'Synthesizing mixtures in parallel across {num_workers} threads')
for key, params in mixture_parameters.items():
synthesize_mixtures_in_parallel(params['target_path'], num_workers)
def test_polyphony_gini():
'''
Test computation of polyphony gini
'''
# Annotation must have namespace scaper, otherwise raise error
ann = jams.Annotation('tag_open', duration=10)
gini = pytest.raises(ScaperError, polyphony_gini, ann)
# Annotation without duration set should raise error
ann = jams.Annotation('scaper', duration=None)
gini = pytest.raises(ScaperError, polyphony_gini, ann)
# Annotation with no foreground events returns a gini of 0
sc = Scaper(10.0, FG_PATH, BG_PATH)
# add background
sc.add_background(label=('choose', []),
source_file=('choose', []),
source_time=('const', 0))
jam = sc._instantiate()
ann = jam.annotations[0]
gini = polyphony_gini(ann)
assert gini == 0
def __test_gini_from_event_times(event_time_list, expected_gini,
hop_size=0.01):
print(event_time_list)
# create scaper
sc = Scaper(10.0, FG_PATH, BG_PATH)
# add background
sc.add_background(label=('choose', []),
source_file=('choose', []),
source_time=('const', 0))
# add foreground events based on the event time list
# always use siren file since it is 26 s long, so we can choose the
# event duration flexibly
for onset, offset in event_time_list:
sc.add_event(label=('const', 'siren'),
source_file=('const', SIREN_FILE),
source_time=('const', 0),
event_time=('const', onset),
event_duration=('const', offset - onset),
snr=('uniform', 6, 30),
pitch_shift=('uniform', -3, 3),
time_stretch=None)
jam = sc._instantiate()
ann = jam.annotations[0]
gini = polyphony_gini(ann, hop_size=hop_size)
print(gini, expected_gini)
assert np.allclose([gini], [expected_gini], atol=1e-5)
event_time_lists = ([
[],
[(0, 1)],
[(0, 5), (5, 10)],
[(0, 10), (3, 7), (4, 6)]
])
expected_ginis = [0, 0.1, 1, 0.75]
for etl, g in zip(event_time_lists, expected_ginis):
__test_gini_from_event_times(etl, g, hop_size=0.01)
for etl, g in zip(event_time_lists, expected_ginis):
__test_gini_from_event_times(etl, g, hop_size=1.0)