sg*dZddlmZGddeZy)z$Speech processor class for SpeechT5.)ProcessorMixincBeZdZdZdZdZfdZdZdZdZ dZ xZ S) SpeechT5Processora} 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. SpeechT5FeatureExtractorSpeechT5Tokenizerc&t|||y)N)super__init__)selffeature_extractor tokenizer __class__s c/var/www/html/venv/lib/python3.12/site-packages/transformers/models/speecht5/processing_speecht5.pyr zSpeechT5Processor.__init__%s *I6c<|jdd}|jdd}|jdd}|jdd}|jdd}| | td| | td||| | td ||j|g|d|i|}n||j|fi|}nd}||j|||d |} | d } n||j|fi|} | d } nd} || S|  |d <| j d} | | |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. audioNtext text_target audio_target 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.)rr input_values input_idslabelsattention_maskdecoder_attention_mask)pop ValueErrorr r get) r argskwargsrrrrrinputstargetsrrs r__call__zSpeechT5Processor.__call__(s0 7D)zz&$'jj5 zz.$7  ?D9  !1n   # (?n  =\1dl{GZs   +T++E`D` `Y_`F  #T^^D3F3FF  #,d,,,]juntuG^,F  $$dnn[;F;G[)FG >N  %F8 %,[[1A%B "%13I/0 rc|jdd}|jdd}|jdd}| | td|| | td|!|jj|g|i|}n"||jj|fi|}nd}|d|vst |t r*d|dvr#|jj|fi|}|d}nt|jj}|jj|j_|jj|g|i|}||j_|d}nd}||S|||d<|jd} | | |d <|S) au Collates the audio and text inputs, as well as their targets, into a padded batch. Audio inputs are padded by SpeechT5FeatureExtractor's [`~SpeechT5FeatureExtractor.pad`]. Text inputs are padded by SpeechT5Tokenizer's [`~SpeechT5Tokenizer.pad`]. Valid input combinations are: - `input_ids` only - `input_values` only - `labels` only, either log-mel spectrograms or text tokens - `input_ids` and log-mel spectrogram `labels` - `input_values` and text `labels` Please refer to the docstring of the above two methods for more information. rNrrz:Cannot process both `input_values` and `input_ids` inputs.zZYou need to specify either an `input_values`, `input_ids`, or `labels` input to be padded.rr) rrr padr isinstancelist feature_size num_mel_binsr) r rr rrrr!r"feature_size_hackrs rr&zSpeechT5Processor.pados"zz.$7 JJ{D1 Hd+  # (=YZ Z  I$5&.l   #/T++// NtNvNF  "'T^^'' v> -$($:$:$G$G!6:6L6L6Y6Y&&34$0044VMdMfM6G&&3 0G >N  %F8 %,[[1A%B "%13I/0 rc:|jj|i|S)z This method forwards all its arguments to SpeechT5Tokenizer's [`~SpeechT5Tokenizer.batch_decode`]. Please refer to the docstring of this method for more information. )r batch_decoder rr s rr-zSpeechT5Processor.batch_decodes +t~~**D;F;;rc:|jj|i|S)z This method forwards all its arguments to SpeechT5Tokenizer's [`~SpeechT5Tokenizer.decode`]. Please refer to the docstring of this method for more information. )r decoder.s rr0zSpeechT5Processor.decodes %t~~$$d5f55r) __name__ __module__ __qualname____doc__feature_extractor_classtokenizer_classr r#r&r-r0 __classcell__)rs@rrrs1 9)O7EN:x<6rrN)r4processing_utilsrrrrr:s+.c6c6r