MCPcopy Create free account
hub / github.com/espnet/espnet / __init__

Method __init__

espnet2/train/preprocessor.py:139–276  ·  view source on GitHub ↗
(
        self,
        train: bool,
        use_lang_prompt: bool = False,
        use_nlp_prompt: bool = False,
        token_type: Optional[str] = None,
        token_list: Union[Path, str, Iterable[str]] = None,
        bpemodel: Union[Path, str, Iterable[str]] = None,
        text_cleaner: Collection[str] = None,
        g2p_type: Optional[str] = None,
        unk_symbol: str = "<unk>",
        space_symbol: str = "<space>",
        non_linguistic_symbols: Union[Path, str, Iterable[str]] = None,
        delimiter: Optional[str] = None,
        force_single_channel: bool = False,
        rir_scp: Optional[str] = None,
        rir_apply_prob: float = 1.0,
        noise_scp: Optional[str] = None,
        noise_apply_prob: float = 1.0,
        noise_db_range: str = "3_10",
        short_noise_thres: float = 0.5,
        aux_task_names: Collection[str] = None,
        speech_volume_normalize: float = None,
        speech_name: str = "speech",
        text_name: str = "text",
        fs: int = 0,
        nonsplit_symbol: Iterable[str] = None,
        data_aug_effects: List = None,
        data_aug_num: List[int] = [1, 1],
        data_aug_prob: float = 0.0,
        # for padding of chunk iterator, working when > 0
        min_sample_size: int = -1,
        audio_pad_value: Union[float, int] = 0.0,
        # only use for whisper
        whisper_language: Optional[str] = None,
        whisper_task: Optional[str] = None,
    )

Source from the content-addressed store, hash-verified

137
138class CommonPreprocessor(AbsPreprocessor):
139 def __init__(
140 self,
141 train: bool,
142 use_lang_prompt: bool = False,
143 use_nlp_prompt: bool = False,
144 token_type: Optional[str] = None,
145 token_list: Union[Path, str, Iterable[str]] = None,
146 bpemodel: Union[Path, str, Iterable[str]] = None,
147 text_cleaner: Collection[str] = None,
148 g2p_type: Optional[str] = None,
149 unk_symbol: str = "<unk>",
150 space_symbol: str = "<space>",
151 non_linguistic_symbols: Union[Path, str, Iterable[str]] = None,
152 delimiter: Optional[str] = None,
153 force_single_channel: bool = False,
154 rir_scp: Optional[str] = None,
155 rir_apply_prob: float = 1.0,
156 noise_scp: Optional[str] = None,
157 noise_apply_prob: float = 1.0,
158 noise_db_range: str = "3_10",
159 short_noise_thres: float = 0.5,
160 aux_task_names: Collection[str] = None,
161 speech_volume_normalize: float = None,
162 speech_name: str = "speech",
163 text_name: str = "text",
164 fs: int = 0,
165 nonsplit_symbol: Iterable[str] = None,
166 data_aug_effects: List = None,
167 data_aug_num: List[int] = [1, 1],
168 data_aug_prob: float = 0.0,
169 # for padding of chunk iterator, working when > 0
170 min_sample_size: int = -1,
171 audio_pad_value: Union[float, int] = 0.0,
172 # only use for whisper
173 whisper_language: Optional[str] = None,
174 whisper_task: Optional[str] = None,
175 ):
176 super().__init__(train)
177 self.train = train
178 self.speech_name = speech_name
179 self.text_name = text_name
180 self.speech_volume_normalize = speech_volume_normalize
181 self.force_single_channel = force_single_channel
182 self.rir_apply_prob = rir_apply_prob
183 self.noise_apply_prob = noise_apply_prob
184 self.short_noise_thres = short_noise_thres
185 self.aux_task_names = aux_task_names
186 self.use_lang_prompt = use_lang_prompt
187 self.use_nlp_prompt = use_nlp_prompt
188
189 if token_type is not None:
190 if token_list is None:
191 raise ValueError("token_list is required if token_type is not None")
192 self.text_cleaner = TextCleaner(text_cleaner)
193
194 self.tokenizer = build_tokenizer(
195 token_type=token_type,
196 bpemodel=bpemodel,

Callers

nothing calls this directly

Calls 9

TextCleanerClass · 0.90
build_tokenizerFunction · 0.90
TokenIDConverterClass · 0.90
DataAugmentationClass · 0.90
appendMethod · 0.80
__init__Method · 0.45
splitMethod · 0.45

Tested by

no test coverage detected