def parser():
_parser = config_argparse.ArgumentParser("test")
_parser.add_argument("--foo")
_parser.add_argument("--bar")
_parser.add_argument("--baz", action="store_true")
_parser.add_argument("--count", action="count")
return _parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="Frontend inference",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--ref_scp",
type=str,
required=True,
action="append",
)
group.add_argument(
"--inf_scp",
type=str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str)
group.add_argument("--metrics", type=str, action="append")
group.add_argument("--ref_channel", type=int, default=0)
group.add_argument(
"--frame_size",
type=int,
default=512,
help="STFT frame size in samples, for calculating framewise-* metrics",
)
group.add_argument(
"--frame_hop",
type=int,
default=256,
help="STFT frame hop in samples, for calculating framewise-* metrics",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="Upload files to Zenodo",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
parser.add_argument(
"--access_token",
help="Get your access_token from "
"https://zenodo.org/account/settings/applications/ or "
"https://sandbox.zenodo.org/account/settings/applications/ . "
"You can also give it from an environment variable 'ACCESS_TOKEN'",
)
parser.add_argument(
"--title",
required=True,
help="e.g. ESPnet pretrained model, MT, "
"Fisher-CallHome Spanish (Es->En), Transformer",
)
parser.add_argument("--creator_name", required=True, help="Your name")
parser.add_argument("--file", nargs="+", required=True)
group = parser.add_mutually_exclusive_group(required=True)
group.add_argument("--description", help="Give the description")
group.add_argument("--description_file",
help="Give the description from file")
parser.add_argument(
"--use_sandbox",
type=str2bool,
default=False,
help="Use zenodo sandbox for testing",
)
parser.add_argument("--publish",
type=str2bool,
default=False,
help="Publish after uploading")
parser.add_argument("--license", default="CC-BY-4.0")
parser.add_argument("--affiliation")
parser.add_argument("--orcid")
parser.add_argument("--gnd")
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="Frontend inference",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument("--fs",
type=humanfriendly_or_none,
default=8000,
help="Sampling rate")
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("Output data related")
group.add_argument(
"--normalize_output_wav",
type=str2bool,
default=False,
help="Whether to normalize the predicted wav to [-1~1]",
)
group = parser.add_argument_group("The model configuration related")
group.add_argument(
"--train_config",
type=str,
help="Training configuration file",
)
group.add_argument(
"--model_file",
type=str,
help="Model parameter file",
)
group.add_argument(
"--model_tag",
type=str,
help="Pretrained model tag. If specify this option, train_config and "
"model_file will be overwritten",
)
group.add_argument(
"--inference_config",
type=str_or_none,
default=None,
help="Optional configuration file for overwriting enh model attributes "
"during inference",
)
group.add_argument(
"--enh_s2t_task",
type=str2bool,
default=False,
help="enhancement and asr joint model",
)
group = parser.add_argument_group("Data loading related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group = parser.add_argument_group("SeparateSpeech related")
group.add_argument(
"--segment_size",
type=float,
default=None,
help=
"Segment length in seconds for segment-wise speech enhancement/separation",
)
group.add_argument(
"--hop_size",
type=float,
default=None,
help=
"Hop length in seconds for segment-wise speech enhancement/separation",
)
group.add_argument(
"--normalize_segment_scale",
type=str2bool,
default=False,
help=
"Whether to normalize the energy of the separated streams in each segment",
)
group.add_argument(
"--show_progressbar",
type=str2bool,
default=False,
help=
"Whether to show a progress bar when performing segment-wise speech "
"enhancement/separation",
)
group.add_argument(
"--ref_channel",
type=int,
default=None,
help="If not None, this will overwrite the ref_channel defined in the "
"separator module (for multi-channel speech processing)",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="ASR Decoding",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("INFO", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument("--asr_train_config", type=str, required=True)
group.add_argument("--asr_model_file", type=str, required=True)
group.add_argument("--lm_train_config", type=str)
group.add_argument("--lm_file", type=str)
group.add_argument("--word_lm_train_config", type=str)
group.add_argument("--word_lm_file", type=str)
group = parser.add_argument_group("Beam-search related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group.add_argument("--nbest",
type=int,
default=1,
help="Output N-best hypotheses")
group.add_argument("--beam_size", type=int, default=20, help="Beam size")
group.add_argument("--penalty",
type=float,
default=0.0,
help="Insertion penalty")
group.add_argument(
"--maxlenratio",
type=float,
default=0.0,
help="Input length ratio to obtain max output length. "
"If maxlenratio=0.0 (default), it uses a end-detect "
"function "
"to automatically find maximum hypothesis lengths",
)
group.add_argument(
"--minlenratio",
type=float,
default=0.0,
help="Input length ratio to obtain min output length",
)
group.add_argument(
"--ctc_weight",
type=float,
default=0.5,
help="CTC weight in joint decoding",
)
group.add_argument("--lm_weight",
type=float,
default=1.0,
help="RNNLM weight")
group.add_argument(
"--blank_symbol",
type=str,
default="<blank>",
help="The token symbol represents CTC-blank",
)
group = parser.add_argument_group("Text converter related")
group.add_argument(
"--token_type",
type=str_or_none,
default=None,
choices=["char", "bpe", None],
help="The token type for ASR model. "
"If not given, refers from the training args",
)
group.add_argument(
"--bpemodel",
type=str_or_none,
default=None,
help="The model path of sentencepiece. "
"If not given, refers from the training args",
)
return parser
def get_parser():
"""Obtain an argument-parser for the script interface."""
parser = config_argparse.ArgumentParser(
description="ASR Decoding",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
group = parser.add_argument_group("Model configuration related")
group.add_argument("--asr_train_config", type=str, required=True)
group.add_argument("--asr_model_file", type=str, required=True)
group = parser.add_argument_group("Text converter related")
group.add_argument(
"--token_type",
type=str_or_none,
default=None,
choices=["char", "bpe", None],
help="The token type for ASR model. "
"If not given, refers from the training args",
)
group.add_argument(
"--bpemodel",
type=str_or_none,
default=None,
help="The model path of sentencepiece. "
"If not given, refers from the training args",
)
group = parser.add_argument_group("CTC segmentation related")
group.add_argument(
"--fs",
type=int,
default=16000,
help="Sampling Frequency."
" The sampling frequency (in Hz) is needed to correctly determine the"
" starting and ending time of aligned segments.",
)
group.add_argument(
"--min_window_size",
type=int,
default=None,
help="Minimum window size considered for utterance.",
)
group.add_argument(
"--max_window_size",
type=int,
default=None,
help="Maximum window size considered for utterance.",
)
group.add_argument(
"--set_blank",
type=int,
default=None,
help="Index of model dictionary for blank token.",
)
group.add_argument(
"--gratis_blank",
type=str2bool,
default=False,
help="Set the transition cost of the blank token to zero. Audio sections"
" labeled with blank tokens can then be skipped without penalty. Useful"
" if there are unrelated audio segments between utterances.",
)
group.add_argument(
"--replace_spaces_with_blanks",
type=str2bool,
default=False,
help="Fill blanks in between words to better model pauses between words."
" This option is only active for `--text_converter classic`."
" Segments can be misaligned if this option is combined with"
" --gratis-blank.",
)
group.add_argument(
"--scoring_length",
type=int,
default=None,
help=
"Changes partitioning length L for calculation of the confidence score.",
)
group.add_argument(
"--time_stamps",
type=str,
default=CTCSegmentation.time_stamps,
choices=CTCSegmentation.choices_time_stamps,
help="Select method how CTC index duration is estimated, and"
" thus how the time stamps are calculated.",
)
group.add_argument(
"--text_converter",
type=str,
default=CTCSegmentation.text_converter,
choices=CTCSegmentation.choices_text_converter,
help="How CTC segmentation handles text.",
)
group = parser.add_argument_group("Input/output arguments")
group.add_argument(
"--kaldi_style_text",
type=str2bool,
default=True,
help=
"Assume that the input text file is kaldi-style formatted, i.e., the"
" utterance name is at the beginning of each line.",
)
group.add_argument(
"--print_utt_text",
type=str2bool,
default=True,
help="Include the utterance text in the segments output.",
)
group.add_argument(
"--print_utt_score",
type=str2bool,
default=True,
help="Include the confidence score in the segments output.",
)
group.add_argument(
"-a",
"--audio",
type=Path,
required=True,
help="Input audio file.",
)
group.add_argument(
"-t",
"--text",
type=argparse.FileType("r"),
required=True,
help="Input text file."
" Each line contains the ground truth of a single utterance."
" Kaldi-style text files include the name of the utterance as"
" the first word in the line.",
)
group.add_argument(
"-o",
"--output",
type=argparse.FileType("w"),
default="-",
help="Output in the form of a `segments` file."
" If not given, output is written to stdout.",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="Speaker Diarization inference",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--fs",
type=humanfriendly_parse_size_or_none,
default=8000,
help="Sampling rate",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument(
"--train_config",
type=str,
help="Diarization training configuration",
)
group.add_argument(
"--model_file",
type=str,
help="Diarization model parameter file",
)
group.add_argument(
"--model_tag",
type=str,
help="Pretrained model tag. If specify this option, train_config and "
"model_file will be overwritten",
)
group = parser.add_argument_group("Data loading related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group = parser.add_argument_group("Diarize speech related")
group.add_argument(
"--segment_size",
type=float,
default=None,
help="Segment length in seconds for segment-wise speaker diarization",
)
group.add_argument(
"--show_progressbar",
type=str2bool,
default=False,
help=
"Whether to show a progress bar when performing segment-wise speaker "
"diarization",
)
return parser
def get_parser():
"""Get argument parser."""
parser = config_argparse.ArgumentParser(
description="TTS Decode",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use "_" instead of "-" as separator.
# "-" is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument(
"--output_dir",
type=str,
required=True,
help="The path of output directory",
)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument(
"--seed",
type=int,
default=0,
help="Random seed",
)
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
parser.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument(
"--key_file",
type=str_or_none,
)
group.add_argument(
"--allow_variable_data_keys",
type=str2bool,
default=False,
)
group = parser.add_argument_group("The model configuration related")
group.add_argument(
"--train_config",
type=str,
help="Training configuration file.",
)
group.add_argument(
"--model_file",
type=str,
help="Model parameter file.",
)
group = parser.add_argument_group("Decoding related")
group.add_argument(
"--maxlenratio",
type=float,
default=10.0,
help="Maximum length ratio in decoding",
)
group.add_argument(
"--minlenratio",
type=float,
default=0.0,
help="Minimum length ratio in decoding",
)
group.add_argument(
"--threshold",
type=float,
default=0.5,
help="Threshold value in decoding",
)
group.add_argument(
"--use_att_constraint",
type=str2bool,
default=False,
help="Whether to use attention constraint",
)
group.add_argument(
"--backward_window",
type=int,
default=1,
help="Backward window value in attention constraint",
)
group.add_argument(
"--forward_window",
type=int,
default=3,
help="Forward window value in attention constraint",
)
group.add_argument(
"--use_teacher_forcing",
type=str2bool,
default=False,
help="Whether to use teacher forcing",
)
parser.add_argument(
"--speed_control_alpha",
type=float,
default=1.0,
help="Alpha in FastSpeech to change the speed of generated speech",
)
group = parser.add_argument_group("Grriffin-Lim related")
group.add_argument(
"--vocoder_conf",
action=NestedDictAction,
default=get_default_kwargs(Spectrogram2Waveform),
help="The configuration for Grriffin-Lim",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="ASR Decoding",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument("--asr_train_config", type=str, required=True)
group.add_argument("--asr_model_file", type=str, required=True)
group.add_argument(
"--model_tag",
type=str,
help="Pretrained model tag. If specify this option, *_train_config and "
"*_file will be overwritten",
)
group = parser.add_argument_group("Decoding related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group.add_argument("--maskctc_n_iterations", type=int, default=10)
group.add_argument("--maskctc_threshold_probability",
type=float,
default=0.99)
group = parser.add_argument_group("Text converter related")
group.add_argument(
"--token_type",
type=str_or_none,
default=None,
choices=["char", "bpe", None],
help="The token type for ASR model. "
"If not given, refers from the training args",
)
group.add_argument(
"--bpemodel",
type=str_or_none,
default=None,
help="The model path of sentencepiece. "
"If not given, refers from the training args",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="Calc perplexity",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("INFO", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
parser.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
parser.add_argument(
"--log_base",
type=float_or_none,
default=None,
help="The base of logarithm for Perplexity. "
"If None, napier's constant is used.",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys", type=str2bool, default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument("--train_config", type=str)
group.add_argument("--model_file", type=str)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="ASR Decoding",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument(
"--asr_train_config",
type=str,
help="ASR training configuration",
)
group.add_argument(
"--asr_model_file",
type=str,
help="ASR model parameter file",
)
group.add_argument(
"--lm_train_config",
type=str,
help="LM training configuration",
)
group.add_argument(
"--lm_file",
type=str,
help="LM parameter file",
)
group.add_argument(
"--word_lm_train_config",
type=str,
help="Word LM training configuration",
)
group.add_argument(
"--word_lm_file",
type=str,
help="Word LM parameter file",
)
group.add_argument(
"--ngram_file",
type=str,
help="N-gram parameter file",
)
group.add_argument(
"--model_tag",
type=str,
help="Pretrained model tag. If specify this option, *_train_config and "
"*_file will be overwritten",
)
group.add_argument(
"--enh_s2t_task",
type=str2bool,
default=False,
help="enhancement and asr joint model",
)
group = parser.add_argument_group("Quantization related")
group.add_argument(
"--quantize_asr_model",
type=str2bool,
default=False,
help="Apply dynamic quantization to ASR model.",
)
group.add_argument(
"--quantize_lm",
type=str2bool,
default=False,
help="Apply dynamic quantization to LM.",
)
group.add_argument(
"--quantize_modules",
type=str,
nargs="*",
default=["Linear"],
help="""List of modules to be dynamically quantized.
E.g.: --quantize_modules=[Linear,LSTM,GRU].
Each specified module should be an attribute of 'torch.nn', e.g.:
torch.nn.Linear, torch.nn.LSTM, torch.nn.GRU, ...""",
)
group.add_argument(
"--quantize_dtype",
type=str,
default="qint8",
choices=["float16", "qint8"],
help="Dtype for dynamic quantization.",
)
group = parser.add_argument_group("Beam-search related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group.add_argument("--nbest",
type=int,
default=1,
help="Output N-best hypotheses")
group.add_argument("--beam_size", type=int, default=20, help="Beam size")
group.add_argument("--penalty",
type=float,
default=0.0,
help="Insertion penalty")
group.add_argument(
"--maxlenratio",
type=float,
default=0.0,
help="Input length ratio to obtain max output length. "
"If maxlenratio=0.0 (default), it uses a end-detect "
"function "
"to automatically find maximum hypothesis lengths."
"If maxlenratio<0.0, its absolute value is interpreted"
"as a constant max output length",
)
group.add_argument(
"--minlenratio",
type=float,
default=0.0,
help="Input length ratio to obtain min output length",
)
group.add_argument(
"--ctc_weight",
type=float,
default=0.5,
help="CTC weight in joint decoding",
)
group.add_argument("--lm_weight",
type=float,
default=1.0,
help="RNNLM weight")
group.add_argument("--ngram_weight",
type=float,
default=0.9,
help="ngram weight")
group.add_argument("--streaming", type=str2bool, default=False)
group.add_argument(
"--transducer_conf",
default=None,
help="The keyword arguments for transducer beam search.",
)
group = parser.add_argument_group("Text converter related")
group.add_argument(
"--token_type",
type=str_or_none,
default=None,
choices=["char", "bpe", None],
help="The token type for ASR model. "
"If not given, refers from the training args",
)
group.add_argument(
"--bpemodel",
type=str_or_none,
default=None,
help="The model path of sentencepiece. "
"If not given, refers from the training args",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="Frontend inference",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("INFO", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument("--fs", type=int, default=8000, help="Sampling rate")
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("Output data related")
group.add_argument(
"--normalize_output_wav",
type=str2bool,
default=False,
help="Weather to normalize the predicted wav to [-1~1]",
)
group = parser.add_argument_group("The model configuration related")
group.add_argument("--enh_train_config", type=str, required=True)
group.add_argument("--enh_model_file", type=str, required=True)
group = parser.add_argument_group("Beam-search related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
return parser
def get_parser():
parser = config_argparse.ArgumentParser(
description="ASR Decoding",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
# Note(kamo): Use '_' instead of '-' as separator.
# '-' is confusing if written in yaml.
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys", type=str2bool, default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument(
"--asr_train_config",
type=str,
help="ASR training configuration",
)
group.add_argument(
"--asr_model_file",
type=str,
help="ASR model parameter file",
)
group.add_argument(
"--lm_train_config",
type=str,
help="LM training configuration",
)
group.add_argument(
"--lm_file",
type=str,
help="LM parameter file",
)
group.add_argument(
"--word_lm_train_config",
type=str,
help="Word LM training configuration",
)
group.add_argument(
"--word_lm_file",
type=str,
help="Word LM parameter file",
)
group.add_argument(
"--model_tag",
type=str,
help="Pretrained model tag. If specify this option, *_train_config and "
"*_file will be overwritten",
)
group = parser.add_argument_group("Beam-search related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group.add_argument("--nbest", type=int, default=1, help="Output N-best hypotheses")
group.add_argument("--beam_size", type=int, default=20, help="Beam size")
group.add_argument("--penalty", type=float, default=0.0, help="Insertion penalty")
group.add_argument(
"--maxlenratio",
type=float,
default=0.0,
help="Input length ratio to obtain max output length. "
"If maxlenratio=0.0 (default), it uses a end-detect "
"function "
"to automatically find maximum hypothesis lengths",
)
group.add_argument(
"--minlenratio",
type=float,
default=0.0,
help="Input length ratio to obtain min output length",
)
group.add_argument(
"--ctc_weight",
type=float,
default=0.5,
help="CTC weight in joint decoding",
)
group.add_argument("--lm_weight", type=float, default=1.0, help="RNNLM weight")
group.add_argument("--streaming", type=str2bool, default=False)
group = parser.add_argument_group("Text converter related")
group.add_argument(
"--token_type",
type=str_or_none,
default=None,
choices=["char", "bpe", None],
help="The token type for ASR model. "
"If not given, refers from the training args",
)
group.add_argument(
"--bpemodel",
type=str_or_none,
default=None,
help="The model path of sentencepiece. "
"If not given, refers from the training args",
)
group.add_argument(
"--is_ctc_decoding",
type=str2bool,
default=True,
help="Use ctc topology as decoding graph",
)
group.add_argument("--use_nbest_rescoring", type=str2bool, default=False)
group.add_argument(
"--num_paths",
type=int,
default=1000,
help="The third argument for k2.random_paths",
)
group.add_argument(
"--nbest_batch_size",
type=int,
default=500,
help="batchify nbest list when computing am/lm scores to avoid OOM",
)
group.add_argument(
"--nll_batch_size",
type=int,
default=100,
help="batch_size when computing nll during nbest rescoring",
)
group.add_argument("--k2_config", type=str, help="Config file for decoding with k2")
return parser
def get_parser():
"""Get Transducer model inference parser."""
parser = config_argparse.ArgumentParser(
description="ASR Transducer Decoding",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
parser.add_argument(
"--log_level",
type=lambda x: x.upper(),
default="INFO",
choices=("CRITICAL", "ERROR", "WARNING", "INFO", "DEBUG", "NOTSET"),
help="The verbose level of logging",
)
parser.add_argument("--output_dir", type=str, required=True)
parser.add_argument(
"--ngpu",
type=int,
default=0,
help="The number of gpus. 0 indicates CPU mode",
)
parser.add_argument("--seed", type=int, default=0, help="Random seed")
parser.add_argument(
"--dtype",
default="float32",
choices=["float16", "float32", "float64"],
help="Data type",
)
parser.add_argument(
"--num_workers",
type=int,
default=1,
help="The number of workers used for DataLoader",
)
group = parser.add_argument_group("Input data related")
group.add_argument(
"--data_path_and_name_and_type",
type=str2triple_str,
required=True,
action="append",
)
group.add_argument("--key_file", type=str_or_none)
group.add_argument("--allow_variable_data_keys",
type=str2bool,
default=False)
group = parser.add_argument_group("The model configuration related")
group.add_argument(
"--asr_train_config",
type=str,
help="ASR training configuration",
)
group.add_argument(
"--asr_model_file",
type=str,
help="ASR model parameter file",
)
group.add_argument(
"--lm_train_config",
type=str,
help="LM training configuration",
)
group.add_argument(
"--lm_file",
type=str,
help="LM parameter file",
)
group.add_argument(
"--model_tag",
type=str,
help="Pretrained model tag. If specify this option, *_train_config and "
"*_file will be overwritten",
)
group = parser.add_argument_group("Beam-search related")
group.add_argument(
"--batch_size",
type=int,
default=1,
help="The batch size for inference",
)
group.add_argument("--nbest",
type=int,
default=1,
help="Output N-best hypotheses")
group.add_argument("--beam_size", type=int, default=5, help="Beam size")
group.add_argument("--lm_weight",
type=float,
default=1.0,
help="RNNLM weight")
group.add_argument(
"--beam_search_config",
default={},
help="The keyword arguments for transducer beam search.",
)
group = parser.add_argument_group("Text converter related")
group.add_argument(
"--token_type",
type=str_or_none,
default=None,
choices=["char", "bpe", None],
help="The token type for ASR model. "
"If not given, refers from the training args",
)
group.add_argument(
"--bpemodel",
type=str_or_none,
default=None,
help="The model path of sentencepiece. "
"If not given, refers from the training args",
)
group = parser.add_argument_group("Dynamic quantization related")
parser.add_argument(
"--quantize_asr_model",
type=bool,
default=False,
help="Apply dynamic quantization to ASR model.",
)
parser.add_argument(
"--quantize_modules",
nargs="*",
default=None,
help="""Module names to apply dynamic quantization on.
The module names are provided as a list, where each name is separated
by a comma (e.g.: --quantize-config=[Linear,LSTM,GRU]).
Each specified name should be an attribute of 'torch.nn', e.g.:
torch.nn.Linear, torch.nn.LSTM, torch.nn.GRU, ...""",
)
parser.add_argument(
"--quantize_dtype",
type=str,
default="qint8",
choices=["float16", "qint8"],
help="Dtype for dynamic quantization.",
)
group = parser.add_argument_group("Streaming related")
parser.add_argument(
"--streaming",
type=bool,
default=False,
help="Whether to perform chunk-by-chunk inference.",
)
parser.add_argument(
"--chunk_size",
type=int,
default=16,
help="Number of frames in chunk AFTER subsampling.",
)
parser.add_argument(
"--left_context",
type=int,
default=32,
help="Number of frames in left context of the chunk AFTER subsampling.",
)
parser.add_argument(
"--right_context",
type=int,
default=0,
help=
"Number of frames in right context of the chunk AFTER subsampling.",
)
parser.add_argument(
"--display_partial_hypotheses",
type=bool,
default=False,
help=
"Whether to display partial hypotheses during chunk-by-chunk inference.",
)
return parser
