def main():
mp.freeze_support()
args = parser.parse_args()
fix_path()
if args.subcommand == 'align':
run_align_corpus(args, acoustic_languages)
elif args.subcommand == 'train':
run_train_corpus(args)
elif args.subcommand == 'g2p':
run_g2p(args, g2p_languages)
elif args.subcommand == 'train_g2p':
run_train_g2p(args)
elif args.subcommand == 'validate':
run_validate_corpus(args)
elif args.subcommand == 'download':
run_download(args)
elif args.subcommand == 'train_lm':
run_train_lm(args)
elif args.subcommand == 'train_ivector':
run_train_ivector_extractor(args)
elif args.subcommand == 'annotator':
run_annotator(args)
elif args.subcommand == 'thirdparty':
run_thirdparty(args)
elif args.subcommand == 'transcribe':
run_transcribe_corpus(args)
unfix_path()
def test_generate_pretrained_threshold(
english_g2p_model, basic_corpus_dir, temp_dir, generated_dir
):
if G2P_DISABLED:
pytest.skip("No Pynini found")
output_path = os.path.join(generated_dir, "g2p_out.txt")
command = [
"g2p",
english_g2p_model,
basic_corpus_dir,
output_path,
"-t",
temp_dir,
"-q",
"--clean",
"--g2p_threshold",
"0.95",
]
args, unknown = parser.parse_known_args(command)
run_g2p(args, unknown)
assert os.path.exists(output_path)
d = MultispeakerDictionary(output_path, temporary_directory=temp_dir)
d.dictionary_setup()
assert len(d.word_mapping(1)) > 0
def test_generate_dict_textgrid(
multilingual_ipa_tg_corpus_dir,
english_g2p_model,
generated_dir,
temp_dir,
g2p_config_path,
):
if G2P_DISABLED:
pytest.skip("No Pynini found")
output_file = os.path.join(generated_dir, "tg_g2pped.dict")
command = [
"g2p",
english_g2p_model,
multilingual_ipa_tg_corpus_dir,
output_file,
"-t",
temp_dir,
"-q",
"--clean",
"--debug",
"--config_path",
g2p_config_path,
]
args, unknown = parser.parse_known_args(command)
run_g2p(args, unknown)
assert os.path.exists(output_file)
d = MultispeakerDictionary(dictionary_path=output_file, temporary_directory=temp_dir)
d.dictionary_setup()
assert len(d.word_mapping()) > 0
def test_generate_dict_text_only(
basic_split_dir,
basic_g2p_model_path,
g2p_basic_output,
temp_dir,
g2p_config_path,
):
if G2P_DISABLED:
pytest.skip("No Pynini found")
text_dir = basic_split_dir[1]
command = [
"g2p",
basic_g2p_model_path,
text_dir,
g2p_basic_output,
"-t",
temp_dir,
"-q",
"--clean",
"--debug",
"--config_path",
g2p_config_path,
]
args, unknown = parser.parse_known_args(command)
run_g2p(args, unknown)
assert os.path.exists(g2p_basic_output)
d = MultispeakerDictionary(dictionary_path=g2p_basic_output, temporary_directory=temp_dir)
d.dictionary_setup()
assert len(d.word_mapping()) > 0
def test_generate_orthography_dict(basic_corpus_dir, orth_sick_output,
temp_dir):
if G2P_DISABLED:
pytest.skip('No Pynini found')
command = [
'g2p', basic_corpus_dir, orth_sick_output, '-t', temp_dir, '-q',
'--clean', '-d'
]
args, unknown = parser.parse_known_args(command)
run_g2p(args)
assert os.path.exists(orth_sick_output)
d = Dictionary(orth_sick_output, temp_dir)
assert len(d.words) > 0
def test_generate_orthography_dict(basic_corpus_dir, orth_sick_output,
temp_dir):
if G2P_DISABLED:
pytest.skip('No Pynini found')
args = G2PDummyArgs()
args.g2p_model_path = None
args.input_path = basic_corpus_dir
args.output_path = orth_sick_output
args.temp_directory = temp_dir
run_g2p(args)
assert os.path.exists(orth_sick_output)
d = Dictionary(orth_sick_output, temp_dir)
assert len(d.words) > 0
def test_generate_orthography_dict(basic_corpus_dir, orth_basic_output, temp_dir):
if G2P_DISABLED:
pytest.skip("No Pynini found")
command = [
"g2p",
basic_corpus_dir,
orth_basic_output,
"-t",
temp_dir,
"-q",
"--clean",
"--debug",
"--use_mp",
"False",
]
args, unknown = parser.parse_known_args(command)
run_g2p(args, unknown)
assert os.path.exists(orth_basic_output)
d = MultispeakerDictionary(dictionary_path=orth_basic_output, temporary_directory=temp_dir)
d.dictionary_setup()
assert len(d.word_mapping()) > 0
def main() -> None:
"""
Main function for the MFA command line interface
"""
check_third_party()
hooks = ExitHooks()
hooks.hook()
atexit.register(hooks.history_save_handler)
from colorama import init
init()
parser = create_parser()
mp.freeze_support()
args, unknown = parser.parse_known_args()
for short in ["-c", "-d"]:
if short in unknown:
print(
f"Due to the number of options that `{short}` could refer to, it is not accepted. "
"Please specify the full argument",
file=sys.stderr,
)
sys.exit(1)
try:
if args.subcommand in ["g2p", "train_g2p"]:
try:
import pynini # noqa
except ImportError:
print(
"There was an issue importing Pynini, please ensure that it is installed. If you are on Windows, "
"please use the Windows Subsystem for Linux to use g2p functionality.",
file=sys.stderr,
)
sys.exit(1)
if args.subcommand == "align":
run_align_corpus(args, unknown)
elif args.subcommand == "adapt":
run_adapt_model(args, unknown)
elif args.subcommand == "train":
run_train_acoustic_model(args, unknown)
elif args.subcommand == "g2p":
run_g2p(args, unknown)
elif args.subcommand == "train_g2p":
run_train_g2p(args, unknown)
elif args.subcommand == "validate":
run_validate_corpus(args, unknown)
elif args.subcommand == "validate_dictionary":
run_validate_dictionary(args, unknown)
elif args.subcommand in ["model", "models"]:
run_model(args)
elif args.subcommand == "train_lm":
run_train_lm(args, unknown)
elif args.subcommand == "train_dictionary":
run_train_dictionary(args, unknown)
elif args.subcommand == "train_ivector":
run_train_ivector_extractor(args, unknown)
elif args.subcommand == "classify_speakers": # pragma: no cover
run_classify_speakers(args, unknown)
elif args.subcommand in ["annotator", "anchor"]:
run_anchor()
elif args.subcommand == "transcribe":
run_transcribe_corpus(args, unknown)
elif args.subcommand == "create_segments":
run_create_segments(args, unknown)
elif args.subcommand == "configure":
update_global_config(args)
global GLOBAL_CONFIG
GLOBAL_CONFIG = load_global_config()
elif args.subcommand == "history":
print_history(args)
elif args.subcommand == "version":
from montreal_forced_aligner.utils import get_mfa_version
print(get_mfa_version())
elif args.subcommand == "thirdparty": # Deprecated command
raise DeprecationWarning(
"Necessary thirdparty executables are now installed via conda. Please refer to the installation docs for the updated commands."
)
elif args.subcommand == "download": # Deprecated command
raise DeprecationWarning(
"Downloading models is now run through the `mfa model download` command, please use that instead."
)
except MFAError as e:
if getattr(args, "debug", False):
raise
print(e, file=sys.stderr)
sys.exit(1)
def main():
mp.freeze_support()
args, unknown = parser.parse_known_args()
fix_path()
if args.subcommand in ['align', 'train', 'train_ivector']:
from montreal_forced_aligner.thirdparty.kaldi import validate_alignment_binaries
if not validate_alignment_binaries():
print(
"There was an issue validating Kaldi binaries, please ensure you've downloaded them via the "
"'mfa thirdparty download' command. See 'mfa thirdparty validate' for more detailed information "
"on why this check failed.")
sys.exit(1)
elif args.subcommand in ['transcribe']:
from montreal_forced_aligner.thirdparty.kaldi import validate_transcribe_binaries
if not validate_transcribe_binaries():
print(
"There was an issue validating Kaldi binaries, please ensure you've downloaded them via the "
"'mfa thirdparty download' command. See 'mfa thirdparty validate' for more detailed information "
"on why this check failed. If you are on MacOS, please note that the thirdparty binaries available "
"via the download command do not contain the transcription ones. To get this functionality working "
"for the time being, please build kaldi locally and follow the instructions for running the "
"'mfa thirdparty kaldi' command.")
sys.exit(1)
elif args.subcommand in ['train_dictionary']:
from montreal_forced_aligner.thirdparty.kaldi import validate_train_dictionary_binaries
if not validate_train_dictionary_binaries():
print(
"There was an issue validating Kaldi binaries, please ensure you've downloaded them via the "
"'mfa thirdparty download' command. See 'mfa thirdparty validate' for more detailed information "
"on why this check failed. If you are on MacOS, please note that the thirdparty binaries available "
"via the download command do not contain the train_dictionary ones. To get this functionality working "
"for the time being, please build kaldi locally and follow the instructions for running the "
"'mfa thirdparty kaldi' command.")
sys.exit(1)
elif args.subcommand in ['g2p', 'train_g2p']:
try:
import pynini
except ImportError:
print(
"There was an issue importing Pynini, please ensure that it is installed. If you are on Windows, "
"please use the Windows Subsystem for Linux to use g2p functionality."
)
sys.exit(1)
if args.subcommand == 'align':
run_align_corpus(args, unknown, acoustic_languages)
elif args.subcommand == 'train':
run_train_corpus(args)
elif args.subcommand == 'g2p':
run_g2p(args, g2p_languages)
elif args.subcommand == 'train_g2p':
run_train_g2p(args)
elif args.subcommand == 'validate':
run_validate_corpus(args)
elif args.subcommand == 'download':
run_download(args)
elif args.subcommand == 'train_lm':
run_train_lm(args)
elif args.subcommand == 'train_dictionary':
run_train_dictionary(args)
elif args.subcommand == 'train_ivector':
run_train_ivector_extractor(args)
elif args.subcommand == 'classify_speakers':
run_classify_speakers(args)
elif args.subcommand == 'annotator':
from montreal_forced_aligner.command_line.annotator import run_annotator
run_annotator(args)
elif args.subcommand == 'thirdparty':
run_thirdparty(args)
elif args.subcommand == 'transcribe':
run_transcribe_corpus(args)
elif args.subcommand == 'create_segments':
run_create_segments(args, unknown)
elif args.subcommand == 'version':
print(__version__)
unfix_path()