def test_validate(self):
ds = resources.create_dataset()
ds.utterances['utt-3'].label_lists[corpus.LL_WORD_TRANSCRIPT][0].value = 'max length here 11'
ds.utterances['utt-4'].label_lists[corpus.LL_WORD_TRANSCRIPT][0].value = 'too long here'
val = validation.UtteranceTranscriptionRatioValidator(10, corpus.LL_WORD_TRANSCRIPT)
result = val.validate(ds)
assert not result.passed
assert len(result.invalid_utterances) == 1
assert 'utt-4' in result.invalid_utterances.keys()
def test_validate(self):
ds = resources.create_dataset()
ds.utterances['utt-3'].set_label_list(
annotations.LabelList.create_single('max length here 11',
idx=corpus.LL_WORD_TRANSCRIPT))
ds.utterances['utt-4'].set_label_list(
annotations.LabelList.create_single('too long here',
idx=corpus.LL_WORD_TRANSCRIPT))
val = validation.UtteranceTranscriptionRatioValidator(
10, corpus.LL_WORD_TRANSCRIPT)
result = val.validate(ds)
assert not result.passed
assert len(result.invalid_utterances) == 1
assert 'utt-4' in result.invalid_utterances.keys()
def find_invalid_character_ratios(output_path, corpus):
#
# Find invalid chracter ratios
#
report_path = os.path.join(output_path, 'invalid_character_ratio.json')
if not os.path.isfile(report_path):
print('Validate character ratio ...')
v = validation.UtteranceTranscriptionRatioValidator(
max_characters_per_second=25,
label_list_idx=audiomate.corpus.LL_WORD_TRANSCRIPT,
num_threads=4)
result = v.validate(corpus)
invalid_utts = result.invalid_items
write_report(report_path, invalid_utts)
else:
invalid_utts = read_report(report_path)
print('Validate character ratio - Already done')
return invalid_utts.keys()
def test_name(self):
val = validation.UtteranceTranscriptionRatioValidator(10, corpus.LL_WORD_TRANSCRIPT)
assert val.name() == 'Utterance-Transcription-Ratio ({})'.format(corpus.LL_WORD_TRANSCRIPT)