o ZŽh¬ã@s*dZddlmZGdd„deƒZdgZdS)z$Speech processor class for SpeechT5.é)ÚProcessorMixincsHeZdZdZdZdZ‡fdd„Zdd„Zdd „Zd d „Z d d „Z ‡Z S)ÚSpeechT5Processora} Constructs a SpeechT5 processor which wraps a feature extractor and a tokenizer into a single processor. [`SpeechT5Processor`] offers all the functionalities of [`SpeechT5FeatureExtractor`] and [`SpeechT5Tokenizer`]. See the docstring of [`~SpeechT5Processor.__call__`] and [`~SpeechT5Processor.decode`] for more information. Args: feature_extractor (`SpeechT5FeatureExtractor`): An instance of [`SpeechT5FeatureExtractor`]. The feature extractor is a required input. tokenizer (`SpeechT5Tokenizer`): An instance of [`SpeechT5Tokenizer`]. The tokenizer is a required input. ZSpeechT5FeatureExtractorZSpeechT5Tokenizercstƒ ||¡dS)N)ÚsuperÚ__init__)ÚselfÚfeature_extractorÚ tokenizer©Ú __class__©ú_/var/www/auris/lib/python3.10/site-packages/transformers/models/speecht5/processing_speecht5.pyr%szSpeechT5Processor.__init__c Osd| dd¡}| dd¡}| dd¡}| dd¡}| dd¡}|dur*|dur*tdƒ‚|dur6|dur6tdƒ‚|durJ|durJ|durJ|durJtd ƒ‚|dur]|j|g|¢Rd|i|¤Ž}n|durk|j|fi|¤Ž}nd}|dur|j|||d œ|¤Ž} | d } n|dur“|j|fi|¤Ž} | d } nd} |dur›| S| dur°| |d <|  d¡} | dur°| |d<|S)a Processes audio and text input, as well as audio and text targets. You can process audio by using the argument `audio`, or process audio targets by using the argument `audio_target`. This forwards the arguments to SpeechT5FeatureExtractor's [`~SpeechT5FeatureExtractor.__call__`]. You can process text by using the argument `text`, or process text labels by using the argument `text_target`. This forwards the arguments to SpeechT5Tokenizer's [`~SpeechT5Tokenizer.__call__`]. Valid input combinations are: - `text` only - `audio` only - `text_target` only - `audio_target` only - `text` and `audio_target` - `audio` and `audio_target` - `text` and `text_target` - `audio` and `text_target` Please refer to the docstring of the above two methods for more information. ÚaudioNÚtextÚ text_targetÚ audio_targetÚ sampling_ratez\Cannot process both `audio` and `text` inputs. Did you mean `audio_target` or `text_target`?z\Cannot process both `audio_target` and `text_target` inputs. Did you mean `audio` or `text`?zaYou need to specify either an `audio`, `audio_target`, `text`, or `text_target` input to process.)rrÚ input_valuesÚ input_idsÚlabelsÚattention_maskÚdecoder_attention_mask)ÚpopÚ ValueErrorrrÚget) rÚargsÚkwargsr rrrrÚinputsÚtargetsrrr r r Ú__call__(sJ     ÿÿ ÿ   zSpeechT5Processor.__call__c Osd| dd¡}| dd¡}| dd¡}|dur|durtdƒ‚|dur.|dur.|dur.tdƒ‚|dur@|jj|g|¢Ri|¤Ž}n|durO|jj|fi|¤Ž}nd}|dur“d|vsdt|tƒrsd|dvrs|jj|fi|¤Ž}|d}n"|jj}|jj|j_|jj|g|¢Ri|¤Ž}||j_|d}nd}|dur›|S|dur°||d<|  d¡} | dur°| |d <|S) au Collates the audio and text inputs, as well as their targets, into a padded batch. Audio inputs are padded by SpeechT5FeatureExtractor's [`~SpeechT5FeatureExtractor.pad`]. Text inputs are padded by SpeechT5Tokenizer's [`~SpeechT5Tokenizer.pad`]. Valid input combinations are: - `input_ids` only - `input_values` only - `labels` only, either log-mel spectrograms or text tokens - `input_ids` and log-mel spectrogram `labels` - `input_values` and text `labels` Please refer to the docstring of the above two methods for more information. rNrrz:Cannot process both `input_values` and `input_ids` inputs.zZYou need to specify either an `input_values`, `input_ids`, or `labels` input to be padded.érr) rrrÚpadrÚ isinstanceÚlistZ feature_sizeZ num_mel_binsr) rrrrrrrrZfeature_size_hackrr r r r os@   ÿ    zSpeechT5Processor.padcOó|jj|i|¤ŽS)z¿ This method forwards all its arguments to SpeechT5Tokenizer's [`~SpeechT5Tokenizer.batch_decode`]. Please refer to the docstring of this method for more information. )rÚ batch_decode©rrrr r r r$«ózSpeechT5Processor.batch_decodecOr#)z¹ This method forwards all its arguments to SpeechT5Tokenizer's [`~SpeechT5Tokenizer.decode`]. Please refer to the docstring of this method for more information. )rÚdecoder%r r r r'²r&zSpeechT5Processor.decode) Ú__name__Ú __module__Ú __qualname__Ú__doc__Zfeature_extractor_classZtokenizer_classrrr r$r'Ú __classcell__r r r r rs  G<rN)r+Zprocessing_utilsrrÚ__all__r r r r Ús   '