def __init__(
self,
manifest_filepath,
labels,
featurizer,
max_duration=None,
min_duration=None,
max_utts=0,
blank_index=-1,
unk_index=-1,
normalize=True,
trim=False,
bos_id=None,
eos_id=None,
load_audio=True,
):
self.collection = collections.ASRAudioText(
manifests_files=manifest_filepath.split(','),
parser=parsers.ENCharParser(
labels=labels,
unk_id=unk_index,
blank_id=blank_index,
do_normalize=normalize,
),
min_duration=min_duration,
max_duration=max_duration,
max_number=max_utts,
)
self.featurizer = featurizer
self.trim = trim
self.eos_id = eos_id
self.bos_id = bos_id
self.load_audio = load_audio
def __init__(self, punct=True, spaces=False):
labels = []
self.space, labels = len(labels), labels + [' '] # Space
labels.extend(string.ascii_lowercase + "'") # Apostrophe for saving "don't" and "Joe's"
if punct:
labels.extend(self.PUNCT)
super().__init__(labels)
self.punct = punct
self.spaces = spaces
self._parser = parsers.ENCharParser(labels)
def __init__(
self, punct=True, spaces=False, apostrophe=True, add_blank_at="last_but_one",
):
labels = []
self.space, labels = len(labels), labels + [' '] # Space
labels.extend(string.ascii_lowercase)
if apostrophe:
labels.append("'") # Apostrophe for saving "don't" and "Joe's"
if punct:
labels.extend(self.PUNCT)
super().__init__(labels, add_blank_at=add_blank_at)
self.punct = punct
self.spaces = spaces
self._parser = parsers.ENCharParser(labels)