def syllabify_interval(self, phonemes, from_p, to_p, syllables):
"""Perform the syllabification of one interval.
:param phonemes: (sppasTier)
:param from_p: (int) index of the first phoneme to be syllabified
:param to_p: (int) index of the last phoneme to be syllabified
:param syllables: (sppasTier)
"""
# create the sequence of phonemes to syllabify
p = list()
for ann in phonemes[from_p:to_p + 1]:
tag = ann.get_best_tag()
p.append(tag.get_typed_content())
# create the sequence of syllables
s = self.__syllabifier.annotate(p)
# add the syllables into the tier
for i, syll in enumerate(s):
start_idx, end_idx = syll
# create the location
begin = phonemes[start_idx +
from_p].get_lowest_localization().copy()
end = phonemes[end_idx + from_p].get_highest_localization().copy()
location = sppasLocation(sppasInterval(begin, end))
# create the label
syll_string = Syllabifier.phonetize_syllables(p, [syll])
label = sppasLabel(sppasTag(syll_string))
# add the syllable
syllables.create_annotation(location, label)
def __add_repetition(repetition, spk1_tier, spk2_tier, start_idx1,
start_idx2, src_tier, echo_tier):
"""Add a repetition - source and echos - in tiers.
:param repetition: (DataRepetition)
:param spk1_tier: (Tier) The tier of speaker 1 (to detect sources)
:param spk2_tier: (Tier) The tier of speaker 2 (to detect echos)
:param start_idx1: start index of the interval in spk1_tier
:param start_idx2: start index of the interval in spk2_tier
:param src_tier: (Tier) The resulting tier with sources
:param echo_tier: (Tier) The resulting tier with echos
:returns: (bool) the repetition was added or not
"""
index = len(src_tier)
# Source
s, e = repetition.get_source()
src_begin = spk1_tier[start_idx1 + s].get_lowest_localization()
src_end = spk1_tier[start_idx1 + e].get_highest_localization()
time = sppasInterval(src_begin.copy(), src_end.copy())
try:
a = src_tier.create_annotation(
sppasLocation(time),
sppasLabel(sppasTag("S" + str(index + 1))))
src_id = a.get_meta('id')
except TierAddError:
return False
# Echos
for (s, e) in repetition.get_echos():
rep_begin = spk2_tier[start_idx2 + s].get_lowest_localization()
rep_end = spk2_tier[start_idx2 + e].get_highest_localization()
time = sppasInterval(rep_begin.copy(), rep_end.copy())
r = sppasLabel(sppasTag("R" + str(index + 1)))
try:
a = echo_tier.create_annotation(sppasLocation(time), r)
a.set_meta('is_other_repetition_of', src_id)
except TierAddError:
a = echo_tier.find(rep_begin, rep_end)
if len(a) > 0:
a[0].append_label(r)
return True
def __add_repetition(repetition, spk_tier, start_idx, src_tier, echo_tier):
"""Add a repetition - source and echos - in tiers.
:param repetition: (DataRepetition)
:param spk_tier: (sppasTier) The tier of the speaker (to detect sources)
:param start_idx: (int) start index of the interval in spk_tier
:param src_tier: (sppasTier) The resulting tier with sources
:param echo_tier: (sppasTier) The resulting tier with echos
:returns: (bool) the repetition was added or not
"""
index = len(src_tier)
# Source
s, e = repetition.get_source()
src_begin = spk_tier[start_idx + s].get_lowest_localization()
src_end = spk_tier[start_idx + e].get_highest_localization()
time = sppasInterval(src_begin.copy(), src_end.copy())
try:
a = src_tier.create_annotation(
sppasLocation(time),
sppasLabel(sppasTag("S" + str(index + 1))))
src_id = a.get_meta('id')
except:
return False
# Echos
for (s, e) in repetition.get_echos():
rep_begin = spk_tier[start_idx + s].get_lowest_localization()
rep_end = spk_tier[start_idx + e].get_highest_localization()
time = sppasInterval(rep_begin.copy(), rep_end.copy())
a = echo_tier.create_annotation(
sppasLocation(time),
sppasLabel(sppasTag("R" + str(index + 1))))
a.set_meta('is_self_repetition_of', src_id)
return True
b = (i + start) * delta
e = b + delta
for t in tiers_numbers:
tier = trs_input[t - 1]
# get only ONE annotation in our range
anns = tier.find(b, e, overlaps=True)
if len(anns) > 1:
anni = tier.near(b + int(delta / 2.), direction=0)
ann = tier[anni]
else:
ann = anns[0]
texts.append(ann.serialize_labels())
# Append in new tier
ti = sppasInterval(sppasPoint(b, 0.0001), sppasPoint(e, 0.0001))
if len(texts) > 1:
missing = False
for t in texts:
if len(t.strip()) == 0:
# missing annotation label...
missing = True
if missing is True:
text = ""
else:
text = ";".join(texts)
else:
text = str(texts[0])
behavior_tier.create_annotation(sppasLocation(ti),
sppasLabel(sppasTag(text)))
for hyp_ann in hyp_tier:
if hyp_ann in to_merge_anns:
anns_to_merge = to_merge_anns[hyp_ann]
if anns_to_merge is not None:
# a. merge ipus (hyp5)
# ref: # | ipu | #
# hyp: # | ipu | # | ipu | #
nb_hyp_merge_ipus += len(anns_to_merge) - 1
labels = []
for h in anns_to_merge:
labels.extend(h.get_labels())
labels.append(sppasLabel(sppasTag('Merged')))
a = a_hyp_tier.create_annotation(
sppasLocation(
sppasInterval(
to_merge_anns[hyp_ann][0].get_location().get_best().get_begin(),
to_merge_anns[hyp_ann][-1].get_location().get_best().get_end())),
labels
)
else:
a_hyp_tier.add(hyp_ann)
else:
a_hyp_tier = hyp_tier
# Communicate the results:
# ------------------------
trs.append(tier_ref_result)
logging.info(' ==> Match success is {:d} ({:.2f}%)'
''.format(nb_ref_perfect_match,
(float(nb_ref_perfect_match) / float(nb_ipus_ref)) * 100.))
try:
trs_input.shift(args.d)
except AnnDataTypeError:
trs_input.shift(int(args.d))
shifted_trs_start_point = trs_input.get_min_loc().copy()
shifted_trs_end_point = trs_input.get_max_loc().copy()
# ----------------------------------------------------------------------------
# Write
parser.set_filename(args.o)
if sppasRW.create_trs_from_extension(args.o).gaps_support() is False and \
args.nofill is False:
for tier in trs_input:
if tier.is_point():
continue
else:
if args.d > 0:
tier.create_annotation(
sppasLocation(sppasInterval(trs_start_point,
shifted_trs_start_point)))
elif args.d < 0:
tier.create_annotation(
sppasLocation(sppasInterval(shifted_trs_end_point,
trs_end_point)))
if args.quiet is False:
print("Write output: {:s}".format(args.o))
parser.write(trs_input)