o Ni@sdZddlmZddlmZddlmZddlZddlZddlZddl m m Z ddl mZdZdZdZd Zd Zd Zgd Zd dZddZGdddejjZdS)zCREMA-D dataset.)absolute_import)division)print_functionNab @article{cao2014crema, title={{CREMA-D}: Crowd-sourced emotional multimodal actors dataset}, author={Cao, Houwei and Cooper, David G and Keutmann, Michael K and Gur, Ruben C and Nenkova, Ani and Verma, Ragini}, journal={IEEE transactions on affective computing}, volume={5}, number={4}, pages={377--390}, year={2014}, publisher={IEEE} } a CREMA-D is an audio-visual data set for emotion recognition. The data set consists of facial and vocal emotional expressions in sentences spoken in a range of basic emotional states (happy, sad, anger, fear, disgust, and neutral). 7,442 clips of 91 actors with diverse ethnic backgrounds were collected. This release contains only the audio stream from the original audio-visual recording. The samples are splitted between train, validation and testing so that samples from each speaker belongs to exactly one split. z1https://github.com/CheyneyComputerScience/CREMA-DzEhttps://storage.googleapis.com/tfds-data/manual_checksums/crema_d.txtzihttps://raw.githubusercontent.com/CheyneyComputerScience/CREMA-D/master/processedResults/summaryTable.csvzYhttps://media.githubusercontent.com/media/CheyneyComputerScience/CREMA-D/master/AudioWAV/)NEUHAPSADANGFEADIScCst||krtdjt||dtdd|D}td|dkr)td|g}d}|D]\}}|}||7}||t||t||fq/|d d |d d|f|d <|S) aComputes boundary indices for each of the splits in split_probs. Args: split_probs: List of (split_name, prob), e.g. [('train', 0.6), ('dev', 0.2), ('test', 0.2)] n_items: Number of items we want to split. Returns: The item indices of boundaries between different splits. For the above example and n_items=100, these will be [('train', 0, 60), ('dev', 60, 80), ('test', 80, 100)]. z]Not enough items for the splits. There are {splits} splits while there are only {items} items)splitsitemscs|]\}}|VqdSN).0nameprrU/home/ubuntu/.local/lib/python3.10/site-packages/tensorflow_datasets/audio/crema_d.py Nz,_compute_split_boundaries..g:0yE>z"Probs should sum up to 1. probs={}gr)len ValueErrorformatsumabsappendint) split_probsn_items total_probssplit_boundariessum_prrprevrrr_compute_split_boundaries=s$  "r%cCsttdd|D}tj|}||t|t|}i}|D]\}}} t|| D]} |||| <q+q!t t} |D]\} } || }| | | q<| S)aDSplit items to train/dev/test, so all items in group go into same split. Each group contains all the samples from the same speaker ID. The samples are splitted between train, validation and testing so that samples from each speaker belongs to exactly one split. Args: items_and_groups: Sequence of (item_id, group_id) pairs. split_probs: List of (split_name, prob), e.g. [('train', 0.6), ('dev', 0.2), ('test', 0.2)] split_number: Generated splits should change with split_number. Returns: Dictionary that looks like {split name -> set(ids)}. csr rr)ritem_idgroup_idrrrrorz-_get_inter_splits_by_group..) sortedsetnprandom RandomStateshuffler%rrange collections defaultdictadd)items_and_groupsr split_numbergroupsrngr"group_id_to_split split_namei_starti_endi split_to_idsr&r'splitrrr_get_inter_splits_by_group_s    r=c@s4eZdZdZejdZddZddZ ddZ d S) CremaDz:The audio part of CREMA-D dataset for emotion recognition.z1.0.0c CsDtjj|ttjtjjdddtjjtt dt j ddt t dS)Nwavi>) file_format sample_rate)namesaudiolabel speaker_id)rDrE)builder descriptionfeaturessupervised_keyshomepagecitation)tfdscore DatasetInfo _DESCRIPTIONrI FeaturesDictAudio ClassLabellistLABELStfstring _HOMEPAGE _CITATION)selfrrr_infosz CremaD._infocCsb|t|dti}g}g}g}tgd}tjj|d>}|D]3}| dd dd} | r9| |vr:q$t j td| } || || dd || q$Wd n1sbwY|d |i}tt|d |}tt||} gd } t| | d } tjjtjjd | didtjjtjjd | didtjjtjjd | didgS)zReturns SplitGenerators. summary_table)FileName1040_ITH_SAD_XX1006_TIE_NEU_XX1013_WSI_DIS_XX1017_IWW_FEA_XX,r"z%s.wav_rN all_files))traingffffff?) validationg?)testg?file_paths_and_namesrg)r gen_kwargsrhri)download_checksums_CHECKSUMS_URLdownloadSUMMARY_TABLE_URLr)rViogfileGFilestripr<replaceospathjoin WAV_DATA_URLrrTzipr=rMrNSplitGeneratorSplitTRAIN VALIDATIONTEST)rZ dl_managercsv_path all_wav_files speaker_ids wav_names bad_filesflinewav_namewav_pathwav_and_speaker_idsrr rrr_split_generatorssF          zCremaD._split_generatorsccsF|D]\}}|dd}|dd}|||d}||fVqdS)zYields examples.rerrCN)r<)rZrj file_path file_namerFrEexamplerrr_generate_exampless   zCremaD._generate_examplesN) __name__ __module__ __qualname____doc__rMrNVersionVERSIONr[rrrrrrr>s   'r>)r __future__rrrr/runumpyr*tensorflow.compat.v2compatv2rVtensorflow_datasets.public_api public_apirMrYrPrXrmrorxrUr%r=rNGeneratorBasedBuilderr>rrrrs&      ""