a h @sdZddlZddlmZddlmZmZddlmZm Z m Z ddl m Z m Z mZdd lmZdd lmZGd d d ed dZGddde ZdgZdS)z% Speech processor class for Wav2Vec2 N)contextmanager)OptionalUnion)ProcessingKwargsProcessorMixinUnpack) AudioInputPreTokenizedInput TextInput)Wav2Vec2FeatureExtractor)Wav2Vec2CTCTokenizerc@seZdZiZdS)Wav2Vec2ProcessorKwargsN)__name__ __module__ __qualname__ _defaultsrrl/var/www/html/assistant/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/processing_wav2vec2.pyrsrF)totalcseZdZdZdZdZfddZefddZde e e e e e eefeed d d Zd d ZeddZeddZZS)Wav2Vec2Processora Constructs a Wav2Vec2 processor which wraps a Wav2Vec2 feature extractor and a Wav2Vec2 CTC tokenizer into a single processor. [`Wav2Vec2Processor`] offers all the functionalities of [`Wav2Vec2FeatureExtractor`] and [`PreTrainedTokenizer`]. See the docstring of [`~Wav2Vec2Processor.__call__`] and [`~Wav2Vec2Processor.decode`] for more information. Args: feature_extractor (`Wav2Vec2FeatureExtractor`): An instance of [`Wav2Vec2FeatureExtractor`]. The feature extractor is a required input. tokenizer ([`PreTrainedTokenizer`]): An instance of [`PreTrainedTokenizer`]. The tokenizer is a required input. r Z AutoTokenizercs t|||j|_d|_dS)NF)super__init__feature_extractorcurrent_processor_in_target_context_manager)selfr tokenizer __class__rrr3szWav2Vec2Processor.__init__c sxztj|fi|WSttfyrtd|jdttj|fi|}t j|fi|}|||dYS0dS)NzLoading a tokenizer inside a from a config that does not include a `tokenizer_class` attribute is deprecated and will be removed in v5. Please add `'tokenizer_class': 'Wav2Vec2CTCTokenizer'` attribute to either your `config.json` or `tokenizer_config.json` file to suppress this warning: )rr) rfrom_pretrainedOSError ValueErrorwarningswarnr FutureWarningr r)clsZpretrained_model_name_or_pathkwargsrrrrrr!8s  z!Wav2Vec2Processor.from_pretrainedN)audiotextr(c Ksd|vrtd|d}|dur4|dur4td|jtfd|jji|}|jrz|j |fi|d|d|dS|dur|j |fi|d}|dur|j|fi|d}|dur|S|dur|S|d |d <|SdS) a* This method forwards all arguments to [`Wav2Vec2FeatureExtractor.__call__`] and/or [`PreTrainedTokenizer.__call__`] depending on the input modality and returns their outputs. If both modalities are passed, [`Wav2Vec2FeatureExtractor.__call__`] and [`PreTrainedTokenizer.__call__`] are called. Args: audio (`np.ndarray`, `torch.Tensor`, `List[np.ndarray]`, `List[torch.Tensor]`, *optional*): An audio input is passed to [`Wav2Vec2FeatureExtractor.__call__`]. text (`str`, `List[str]`, *optional*): A text input is passed to [`PreTrainedTokenizer.__call__`]. Returns: This method returns the results of each `call` method. If both are used, the output is a dictionary containing the results of both. Z raw_speechzLUsing `raw_speech` as a keyword argument is deprecated. Use `audio` instead.NzAYou need to specify either an `audio` or `text` input to process.Ztokenizer_init_kwargsZ audio_kwargsZ text_kwargsZ common_kwargs input_idslabels) r$r%popr#Z _merge_kwargsrrZ init_kwargsrrr) rr)r*ZimagesZvideosr(Z output_kwargsinputs encodingsrrr__call__Ks@   zWav2Vec2Processor.__call__cOs|jr|jj|i|S|dd}|dd}t|dkrP|d}|dd}|durr|jj|g|Ri|}|dur|jj|fi|}|dur|S|dur|S|d|d<|SdS)ag This method operates on batches of extracted features and/or tokenized text. It forwards all arguments to [`Wav2Vec2FeatureExtractor.pad`] and/or [`PreTrainedTokenizer.pad`] depending on the input modality and returns their outputs. If both modalities are passed, [`Wav2Vec2FeatureExtractor.pad`] and [`PreTrainedTokenizer.pad`] are called. Args: input_features: When the first argument is a dictionary containing a batch of tensors, or the `input_features` argument is present, it is passed to [`Wav2Vec2FeatureExtractor.pad`]. labels: When the `label` argument is present, it is passed to [`PreTrainedTokenizer.pad`]. Returns: This method returns the results of each `pad` method. If both are used, the output is a dictionary containing the results of both. input_featuresNr,rr r+)rrpadr-lenrr)rargsr(r1r,rrrr2s"     zWav2Vec2Processor.padcCs|jj}|dgS)Nr,)rmodel_input_names)rZfeature_extractor_input_namesrrrr5sz#Wav2Vec2Processor.model_input_namesccs0tdd|_|j|_dV|j|_d|_dS)z Temporarily sets the tokenizer for processing the input. Useful for encoding the labels when fine-tuning Wav2Vec2. z`as_target_processor` is deprecated and will be removed in v5 of Transformers. You can process your labels by using the argument `text` of the regular `__call__` method (either in the same call as your audio inputs, or in a separate call.TNF)r$r%rrrr)rrrras_target_processorsz%Wav2Vec2Processor.as_target_processor)NNNN)rrr__doc__Zfeature_extractor_classZtokenizer_classr classmethodr!r rrstrlistr r rrr0r2propertyr5rr6 __classcell__rrrrr!s(  8% r)r7r$ contextlibrtypingrrZprocessing_utilsrrrZtokenization_utils_baser r r Zfeature_extraction_wav2vec2r Ztokenization_wav2vec2rrr__all__rrrrs