def cut_set_with_mixed_cut(cut1, cut2):
mixed_cut = MixedCut(
id="mixed-cut-id",
tracks=[MixTrack(cut=cut1),
MixTrack(cut=cut2, offset=1.0, snr=10)],
)
return CutSet({cut.id: cut for cut in [cut1, cut2, mixed_cut]})
def split(manifest: Manifest,
num_splits: int,
randomize: bool = False) -> List[Manifest]:
"""Split a manifest into `num_splits` equal parts. The element order can be randomized."""
num_items = len(manifest)
if num_splits > num_items:
raise ValueError(
f"Cannot split manifest into more chunks ({num_splits}) than its number of items {num_items}"
)
chunk_size = int(ceil(num_items / num_splits))
split_indices = [(i * chunk_size, min(num_items, (i + 1) * chunk_size))
for i in range(num_splits)]
def maybe_randomize(items: Iterable[Any]) -> List[Any]:
items = list(items)
if randomize:
random.shuffle(items)
return items
if isinstance(manifest, RecordingSet):
contents = maybe_randomize(manifest.recordings.items())
return [
RecordingSet(recordings=dict(contents[begin:end]))
for begin, end in split_indices
]
if isinstance(manifest, SupervisionSet):
contents = maybe_randomize(manifest.segments.items())
return [
SupervisionSet(segments=dict(contents[begin:end]))
for begin, end in split_indices
]
if isinstance(manifest, FeatureSet):
contents = maybe_randomize(manifest.features)
return [
FeatureSet(features=contents[begin:end],
feature_extractor=manifest.feature_extractor)
for begin, end in split_indices
]
if isinstance(manifest, CutSet):
contents = maybe_randomize(manifest.cuts.items())
return [
CutSet(cuts=dict(contents[begin:end]))
for begin, end in split_indices
]
raise ValueError(f"Unknown type of manifest: {type(manifest)}")
def cuts():
cuts = CutSet.from_file("test/fixtures/libri/cuts.json")
# Concatenate 100 cut sets together, starting with an empty CutSet()
return sum((cuts.modify_ids(lambda cid: cid + str(i)) for i in range(100)),
CutSet())
