def test_extract_and_store_features_from_cut_set(cut_set, executor,
mix_eagerly):
extractor = Fbank()
with TemporaryDirectory() as tmpdir, LilcomFilesWriter(tmpdir) as storage:
with executor() if executor is not None else no_executor() as ex:
cut_set_with_feats = cut_set.compute_and_store_features(
extractor=extractor,
storage=storage,
mix_eagerly=mix_eagerly,
executor=ex)
# The same number of cuts
assert len(cut_set_with_feats) == 2
for orig_cut, feat_cut in zip(cut_set, cut_set_with_feats):
# The ID is retained
assert orig_cut.id == feat_cut.id
# Features were attached
assert feat_cut.has_features
# Recording is retained unless mixing a MixedCut eagerly
should_have_recording = not (mix_eagerly
and isinstance(orig_cut, MixedCut))
assert feat_cut.has_recording == should_have_recording
cuts = list(cut_set_with_feats)
arr = cuts[0].load_features()
assert arr.shape[0] == 100
assert arr.shape[1] == extractor.feature_dim(cuts[0].sampling_rate)
arr = cuts[1].load_features()
assert arr.shape[0] == 300
assert arr.shape[1] == extractor.feature_dim(cuts[0].sampling_rate)
def test_extract_and_store_features_from_cut_set(cut_set, executor, num_jobs,
storage_type, mix_eagerly):
extractor = Fbank()
with TemporaryDirectory() as tmpdir:
cut_set_with_feats = cut_set.compute_and_store_features(
extractor=extractor,
storage_path=tmpdir,
num_jobs=num_jobs,
mix_eagerly=mix_eagerly,
executor=executor() if executor else None,
storage_type=storage_type).sort_by_duration(
) # sort by duration to ensure the same order of cuts
# The same number of cuts
assert len(cut_set_with_feats) == 2
for orig_cut, feat_cut in zip(cut_set, cut_set_with_feats):
# The ID is retained
assert orig_cut.id == feat_cut.id
# Features were attached
assert feat_cut.has_features
# Recording is retained unless mixing a MixedCut eagerly
should_have_recording = not (mix_eagerly
and isinstance(orig_cut, MixedCut))
assert feat_cut.has_recording == should_have_recording
cuts = list(cut_set_with_feats)
arr = cuts[0].load_features()
assert arr.shape[0] == 300
assert arr.shape[1] == extractor.feature_dim(cuts[0].sampling_rate)
arr = cuts[1].load_features()
assert arr.shape[0] == 100
assert arr.shape[1] == extractor.feature_dim(cuts[0].sampling_rate)
def test_extract_and_store_features(cut):
extractor = Fbank()
with TemporaryDirectory() as tmpdir:
cut_with_feats = cut.compute_and_store_features(extractor=extractor, output_dir=tmpdir)
arr = cut_with_feats.load_features()
assert arr.shape[0] == 100
assert arr.shape[1] == extractor.feature_dim(cut.sampling_rate)
def test_extract_and_store_features(cut):
extractor = Fbank(FbankConfig(sampling_rate=8000))
with TemporaryDirectory() as tmpdir, LilcomFilesWriter(tmpdir) as storage:
cut_with_feats = cut.compute_and_store_features(extractor=extractor,
storage=storage)
arr = cut_with_feats.load_features()
assert arr.shape[0] == 100
assert arr.shape[1] == extractor.feature_dim(cut.sampling_rate)
def test_extract_and_store_features_from_mixed_cut(cut, mix_eagerly):
mixed_cut = cut.append(cut)
extractor = Fbank(FbankConfig(sampling_rate=8000))
with TemporaryDirectory() as tmpdir, LilcomFilesWriter(tmpdir) as storage:
cut_with_feats = mixed_cut.compute_and_store_features(
extractor=extractor, storage=storage, mix_eagerly=mix_eagerly)
arr = cut_with_feats.load_features()
assert arr.shape[0] == 200
assert arr.shape[1] == extractor.feature_dim(mixed_cut.sampling_rate)
def test_extract_and_store_features_from_mixed_cut(cut, mix_eagerly):
mixed_cut = cut.append(cut)
extractor = Fbank()
with TemporaryDirectory() as tmpdir:
cut_with_feats = mixed_cut.compute_and_store_features(
extractor=extractor,
output_dir=tmpdir,
mix_eagerly=mix_eagerly
)
arr = cut_with_feats.load_features()
assert arr.shape[0] == 200
assert arr.shape[1] == extractor.feature_dim(mixed_cut.sampling_rate)
def test_extract_features(cut):
extractor = Fbank(FbankConfig(sampling_rate=8000))
arr = cut.compute_features(extractor=extractor)
assert arr.shape[0] == 100
assert arr.shape[1] == extractor.feature_dim(cut.sampling_rate)