def read_files(file_path, corpus):
path = os.path.dirname(file_path)
for file_idx, file_path in textfile.read_key_value_lines(
file_path, separator=' ').items():
corpus.new_file(os.path.join(path, file_path),
file_idx=file_idx,
copy_file=False)
def feature_scp_generator(path):
""" Return a generator over all feature matrices defined in a scp. """
scp_entries = textfile.read_key_value_lines(path, separator=' ')
for utterance_id, rx_specifier in scp_entries.items():
yield utterance_id, KaldiWriter.read_float_matrix(rx_specifier)
def read_transcriptions(text_path, corpus):
transcriptions = textfile.read_key_value_lines(text_path,
separator=' ')
for utt_id, transcription in transcriptions.items():
ll = annotations.LabelList.create_single(
transcription, idx=audiomate.corpus.LL_WORD_TRANSCRIPT)
corpus.utterances[utt_id].set_label_list(ll)
def read_feature_containers(feat_path, corpus):
if os.path.isfile(feat_path):
base_path = os.path.dirname(feat_path)
containers = textfile.read_key_value_lines(feat_path,
separator=' ')
for container_name, container_path in containers.items():
corpus.new_feature_container(container_name,
path=os.path.join(
base_path, container_path))
def test_read_key_value_lines(self):
file_path = os.path.join(os.path.dirname(__file__),
'key_value_file.txt')
expected = {'a': '1', 'b': '2', 'c': '3'}
records = textfile.read_key_value_lines(file_path, separator=' ')
self.assertDictEqual(expected, records)
def read_utt_to_issuer_mapping(utt_issuer_path, corpus):
utt_issuers = {}
if os.path.isfile(utt_issuer_path):
for utt_id, issuer_idx in textfile.read_key_value_lines(utt_issuer_path, separator=' ').items():
if issuer_idx in corpus.issuers.keys():
utt_issuers[utt_id] = corpus.issuers[issuer_idx]
else:
utt_issuers[utt_id] = corpus.new_issuer(issuer_idx=issuer_idx)
return utt_issuers
def read_genders(genders_path, corpus):
if os.path.isfile(genders_path):
speakers = textfile.read_key_value_lines(genders_path, separator=' ')
for speaker_idx, gender_str in speakers.items():
if gender_str == 'm':
gender = issuers.Gender.MALE
else:
gender = issuers.Gender.FEMALE
speaker = issuers.Speaker(speaker_idx, gender=gender)
corpus.import_issuers(speaker)
def parse_prompts(etc_folder):
""" Read prompts and prompts-orignal and return as dictionary (id as key). """
prompts_path = os.path.join(etc_folder, 'PROMPTS')
prompts_orig_path = os.path.join(etc_folder, 'prompts-original')
prompts = textfile.read_key_value_lines(prompts_path, separator=' ')
prompts_orig = textfile.read_key_value_lines(prompts_orig_path, separator=' ')
prompts_key_fixed = {}
for k, v in prompts.items():
parts = k.split('/')
key = k
if len(parts) > 1:
key = parts[-1]
prompts_key_fixed[key] = v
prompts = prompts_key_fixed
return prompts, prompts_orig
