def test_deterministic(self, prep_org_data):
""" Ensures loading and processing utterences from ELAN files is
deterministic.
"""
bkw_org_path = prep_org_data
utterances_1 = elan.utterances_from_dir(bkw_org_path, ["rf", "xv"])
utterances_2 = elan.utterances_from_dir(bkw_org_path, ["rf", "xv"])
assert utterances_1 == utterances_2
utterances_1 = [
utter for utter in utterances_1 if bkw.bkw_filter(utter)
]
utterances_2 = [
utter for utter in utterances_2 if bkw.bkw_filter(utter)
]
assert utterances_1 == utterances_2
utterances_1 = utterance.remove_duplicates(utterances_1)
utterances_2 = utterance.remove_duplicates(utterances_2)
assert utterances_1 == utterances_2
utterances_1 = [
bkw.bkw_label_segmenter.segment_labels(utter)
for utter in utterances_1
]
utterances_2 = [
bkw.bkw_label_segmenter.segment_labels(utter)
for utter in utterances_2
]
assert utterances_1 == utterances_2
utterances_1 = utterance.remove_empty_text(utterances_1)
utterances_2 = utterance.remove_empty_text(utterances_2)
assert utterances_1 == utterances_2
def test_explore_code_switching(self, prep_org_data):
bkw_org_path = prep_org_data
utterances = elan.utterances_from_dir(bkw_org_path, ["rf", "xv"])
utterances = utterance.remove_empty_text(
utterance.remove_duplicates(utterances))
codeswitched_path = tgt_dir / "codeswitched.txt"
bkw.explore_code_switching(utterances, codeswitched_path)
def test_speaker_durations(self, prep_org_data):
bkw_org_path = prep_org_data
utterances = elan.utterances_from_dir(bkw_org_path, ["rf", "xv"])
print(len(utterances))
utterances = utterance.remove_empty_text(utterances)
print(len(utterances))
utterances = utterance.remove_duplicates(utterances)
print(len(utterances))
utterances = [utter for utter in utterances if bkw.bkw_filter(utter)]
print(len(utterances))
utterances = [
utter for utter in utterances if utterance.duration(utter) < 10000
]
total = 0
fmt = "{:20}{:10}"
dur_fmt = "{:<10.3f}"
print(fmt.format("Speaker", "Duration"))
for speaker, duration in sorted(
utterance.speaker_durations(utterances),
key=lambda x: x[1],
reverse=True):
dur_mins = (duration * ureg.milliseconds).to(
ureg.minutes).magnitude
dur_str = dur_fmt.format(dur_mins)
total += dur_mins
print(fmt.format(speaker, dur_str))
print(fmt.format("Total", dur_fmt.format(total)))
def test_remove_duplicates_same_time():
from persephone.utterance import Utterance, remove_duplicates
utter_a1 = Utterance(org_media_path=Path('test.wav'),
org_transcription_path=Path('test.txt'),
prefix='test',
start_time=1,
end_time=2,
text='test text',
speaker='Unit tester')
utter_a2 = Utterance(org_media_path=Path('test.wav'),
org_transcription_path=Path('test.txt'),
prefix='test',
start_time=1,
end_time=2,
text='test text',
speaker='Unit tester')
utter_b = Utterance(org_media_path=Path('testb.wav'),
org_transcription_path=Path('testb.txt'),
prefix='testb',
start_time=1,
end_time=2,
text='testb text',
speaker='Unit tester')
all_utterances = [utter_a1, utter_a2, utter_b]
result = remove_duplicates(all_utterances)
assert result
assert len(result) == 2
assert utter_b in result
assert (utter_a1 in result or utter_a2 in result)
def test_mark_on_rock_rf_xv_duplicate(self, prep_org_data):
mark_on_rock_path = prep_org_data / "Mark on Rock.eaf"
anbuyika_text = (" Anbuyika rudno karudyo mani arriwa::::m"
" arribebmeng Madjinbardi")
xv_utters = elan.utterances_from_eaf(mark_on_rock_path, ["xv"])
rf_utters = elan.utterances_from_eaf(mark_on_rock_path, ["rf"])
xv_rf_utters = elan.utterances_from_eaf(mark_on_rock_path, ["xv", "rf"])
assert self.check_text_in_utters(anbuyika_text, xv_utters)
assert self.check_text_in_utters(anbuyika_text, rf_utters)
assert self.check_text_in_utters(anbuyika_text, xv_rf_utters)
assert not self.check_text_in_utters("some random text", xv_rf_utters)
assert len(xv_utters) == 425
assert len(rf_utters) == 420
assert len(xv_rf_utters) == 845
assert len(utterance.remove_duplicates(xv_rf_utters)) == 476
assert len(utterance.remove_empty_text(
utterance.remove_duplicates(xv_rf_utters))) == 473
def test_utterances_from_dir(self, prep_org_data):
bkw_org_path = prep_org_data
utterances = elan.utterances_from_dir(bkw_org_path, ["xv"])
assert len(utterances) == 1036
assert len(utterance.remove_empty_text(utterances)) == 1035
assert len(utterance.remove_duplicates(utterances)) == 1029
assert len(
utterance.remove_duplicates(
utterance.remove_empty_text(utterances))) == 1028
utterances = elan.utterances_from_dir(bkw_org_path, ["rf"])
assert len(utterances) == 1242
assert len(utterance.remove_empty_text(utterances)) == 631
assert len(utterance.remove_duplicates(utterances)) == 1239
assert len(
utterance.remove_duplicates(
utterance.remove_empty_text(utterances))) == 631
utterances = elan.utterances_from_dir(bkw_org_path, ["rf", "xv"])
assert len(utterances) == 2278
assert len(utterance.remove_empty_text(utterances)) == 1666
assert len(utterance.remove_duplicates(utterances)) == 1899
assert len(
utterance.remove_duplicates(
utterance.remove_empty_text(utterances))) == 1291