def download_model(model_type, language):
if language is None:
print('Available languages to download for {}:'.format(model_type))
for lang in list_available_languages(model_type):
print(lang)
return
if model_type == 'acoustic':
extension = AcousticModel.extension
out_path = get_pretrained_acoustic_path(language)
elif model_type == 'g2p':
extension = G2PModel.extension
out_path = get_pretrained_g2p_path(language)
elif model_type == 'dictionary':
extension = '.dict'
out_path = get_dictionary_path(language)
elif model_type == 'ivector':
extension = IvectorExtractor.extension
out_path = get_pretrained_ivector_path(language)
else:
raise NotImplementedError
url = 'https://github.com/MontrealCorpusTools/mfa-models/raw/master/{}/{}{}'.format(
model_type, language, extension)
r = requests.get(url)
with open(out_path, 'wb') as f:
f.write(r.content)
def validate_args(args, downloaded_acoustic_models, download_dictionaries):
if not os.path.exists(args.corpus_directory):
raise ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory))
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path))
if not os.path.isfile(args.dictionary_path):
raise ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path))
if args.acoustic_model_path.lower() in downloaded_acoustic_models:
args.acoustic_model_path = get_pretrained_acoustic_path(
args.acoustic_model_path.lower())
elif args.acoustic_model_path.lower().endswith(AcousticModel.extension):
if not os.path.exists(args.acoustic_model_path):
raise ArgumentError('The specified model path does not exist: ' +
args.acoustic_model_path)
else:
raise ArgumentError(
'The language \'{}\' is not currently included in the distribution, '
'please align via training or specify one of the following language names: {}.'
.format(args.acoustic_model_path.lower(),
', '.join(downloaded_acoustic_models)))
def validate(args, download_dictionaries=None):
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate_args(args, downloaded_acoustic_models, download_dictionaries,
downloaded_language_models):
if not os.path.exists(args.corpus_directory):
raise ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory))
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path))
if not os.path.isfile(args.dictionary_path):
raise ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path))
if args.corpus_directory == args.output_directory:
raise ArgumentError(
'Corpus directory and output directory cannot be the same folder.')
if args.acoustic_model_path.lower() in downloaded_acoustic_models:
args.acoustic_model_path = get_pretrained_acoustic_path(
args.acoustic_model_path.lower())
elif args.acoustic_model_path.lower().endswith(AcousticModel.extension):
if not os.path.exists(args.acoustic_model_path):
raise ArgumentError('The specified model path does not exist: ' +
args.acoustic_model_path)
else:
raise ArgumentError(
'The acoustic model \'{}\' is not currently downloaded, '
'please download a pretrained model, align via training or specify one of the following language names: {}.'
.format(args.acoustic_model_path.lower(),
', '.join(downloaded_acoustic_models)))
if args.language_model_path.lower() in downloaded_language_models:
args.language_model_path = get_pretrained_language_model_path(
args.language_model_path.lower())
elif args.language_model_path.lower().endswith(LanguageModel.extension) or \
args.language_model_path.lower().endswith(FORMAT):
if not os.path.exists(args.language_model_path):
raise ArgumentError('The specified model path does not exist: ' +
args.language_model_path)
else:
raise ArgumentError(
'The language model \'{}\' is not currently downloaded, '
'please download, train a new model, or specify one of the following language names: {}.'
.format(args.language_model_path.lower(),
', '.join(downloaded_language_models)))
def validate_args(args, download_dictionaries=None):
if args.dictionary_path is not None and args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(args.dictionary_path.lower())
if not args.source_path.endswith('.arpa'):
if not os.path.exists(args.source_path):
raise (ArgumentError('Could not find the corpus directory {}.'.format(args.source_path)))
if not os.path.isdir(args.source_path):
raise (ArgumentError('The specified corpus directory ({}) is not a directory.'.format(args.source_path)))
else:
if not os.path.exists(args.source_path):
raise (ArgumentError('Could not find the source file {}.'.format(args.source_path)))
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError('Could not find the config file {}.'.format(args.config_path)))
if args.model_path and not os.path.exists(args.model_path):
raise (ArgumentError('Could not find the model file {}.'.format(args.model_path)))
def validate_args(args, download_dictionaries):
if args.corpus_directory == args.output_directory:
raise Exception(
'Corpus directory and output directory cannot be the same folder.')
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory)))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate_args(args, download_dictionaries=None):
if args.test_transcriptions and args.ignore_acoustics:
raise ArgumentError(
'Cannot test transcriptions without acoustic feature generation.')
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory)))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate_args(args, download_dictionaries=None):
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError('Could not find the config file {}.'.format(
args.config_path)))
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory)))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
