o j@s<ddlZddlZddlmZddlmZGdddeZdS)N)Dataset) AugmentWAVcsreZdZ      dfdd Zdd Zd d Zd d ZddZddZddZ ddZ ddZ ddZ Z S)EncoderDataset皙?@ FNc s*t||_||_|j|_t||j|_||_||_||_ | |_ | \|_ |_ddt |j D|_d|_d|_|r_|d|_|jrTd|vsNd|vrTt|||_d|vr_|d|_|j rtdtd |td t|jtd |jtd t|j td |j dSdS)a  Args: ap (TTS.tts.utils.AudioProcessor): audio processor object. meta_data (list): list of dataset instances. seq_len (int): voice segment length in seconds. verbose (bool): print diagnostic information. cSi|]\}}||qSr .0ikeyr r F/home/kuhnn/.local/lib/python3.10/site-packages/TTS/encoder/dataset.py (z+EncoderDataset.__init__..Npadditiverirgaussianz > DataLoader initializationz | > Classes per Batch: z | > Number of instances : z | > Sequence length: z | > Num Classes: z | > Classes: )super__init__configitems sample_rateintseq_lennum_utter_per_classapverboseuse_torch_spec_EncoderDataset__parse_itemsclasses enumerateclassname_to_classid augmentatorgaussian_augmentation_configdata_augmentation_prkeysprintlen) selfrr meta_data voice_lennum_classes_in_batchrraugmentation_configr __class__r rr s6     zEncoderDataset.__init__cCs|jj||jjd}|S)N)sr)rload_wavr)r*filenameaudior r rr2=szEncoderDataset.load_wavcsi}jD]}|d}|jj}||vr|||q|g||<qfdd|D}t|}|g}jD]/}|d}jjdkrO|dn|d}||vrXq?|j dj dkrfq?|||dq?||fS) N audio_filecs$i|]\}}t|jkr||qSr )r)r)r kvr*r rrNs$z0EncoderDataset.__parse_items..emotion_encoder emotion_name speaker_namer) wav_file_path class_name) rrclass_name_keyr'appendlistsortmodelr2shaper)r*class_to_uttersitempath_r=r! new_itemsr r8r __parse_itemsAs*      zEncoderDataset.__parse_itemscC t|jSN)r)rr8r r r__len__b zEncoderDataset.__len__cCrIrJ)r)r!r8r r rget_num_classeserLzEncoderDataset.get_num_classescCs|jSrJ)r!r8r r rget_class_listhszEncoderDataset.get_class_listcCs ||_ddt|jD|_dS)NcSrr r r r r rrmrz.EncoderDataset.set_classes..)r!r"r#)r*r!r r r set_classeskszEncoderDataset.set_classescCstdd|jDS)Ncss|] \}}||fVqdSrJr )r c_nc_idr r r psz>EncoderDataset.get_map_classid_to_classname..)dictr#rr8r r rget_map_classid_to_classnameosz+EncoderDataset.get_map_classid_to_classnamecCs |j|SrJ)r)r*idxr r r __getitem__rrLzEncoderDataset.__getitem__c Csg}g}|D]]}|d}|d}|j|}||}td|jd|j} || | |j}|jdurD|jrDt|jkrD|j|}|j sV|j |} | t | n| t || |qt |}t |}||fS)Nr<r=r)r#r2randomrandintrCrr$r& apply_onerrmelspectrogramr?torch FloatTensorstack LongTensor) r*batchlabelsfeatsrE utter_pathr=class_idwavoffsetmelr r r collate_fnus(       zEncoderDataset.collate_fn)rrrFNN)__name__ __module__ __qualname__rr2r rKrMrNrOrTrVrg __classcell__r r r/rr s"3!r)rWr[torch.utils.datarTTS.encoder.utils.generic_utilsrrr r r rs