o i@sjddlZddlmZddlmZmZmZddlZddlm Z ddl m Z ddddd Z Gd d d e Z dS) N)Path)ListTupleUnion)Dataset)_load_waveform mix_clean mix_singlemix_both) sep_clean enh_singleenh_both sep_noisyc @seZdZdZ     ddeeefded ed ed ed ef d dZde ee j e e j ffddZ dede eee effddZdefddZdede ee j e e j ffddZdS)LibriMixa *LibriMix* :cite:`cosentino2020librimix` dataset. Args: root (str or Path): The path where the directory ``Libri2Mix`` or ``Libri3Mix`` is stored. Not the path of those directories. subset (str, optional): The subset to use. Options: [``"train-360"``, ``"train-100"``, ``"dev"``, and ``"test"``] (Default: ``"train-360"``). num_speakers (int, optional): The number of speakers, which determines the directories to traverse. The Dataset will traverse ``s1`` to ``sN`` directories to collect N source audios. (Default: 2) sample_rate (int, optional): Sample rate of audio files. The ``sample_rate`` determines which subdirectory the audio are fetched. If any of the audio has a different sample rate, raises ``ValueError``. Options: [8000, 16000] (Default: 8000) task (str, optional): The task of LibriMix. Options: [``"enh_single"``, ``"enh_both"``, ``"sep_clean"``, ``"sep_noisy"``] (Default: ``"sep_clean"``) mode (str, optional): The mode when creating the mixture. If set to ``"min"``, the lengths of mixture and sources are the minimum length of all sources. If set to ``"max"``, the lengths of mixture and sources are zero padded to the maximum length of all sources. Options: [``"min"``, ``"max"``] (Default: ``"min"``) Note: The LibriMix dataset needs to be manually generated. Please check https://github.com/JorisCos/LibriMix train-360@r minrootsubset num_speakers sample_ratetaskmodecst|d|d|_tj|jstd|jd|dvr'td|d|dkr5|jd ||n|d krC|jd ||ntd |d||_||_t ||_ |d krcdg|_ n fddt |D|_ dd|j dD|_|jdS)NLibriMixz The path ze doesn't exist. Please check the ``root`` path and ``num_speakers`` or download the dataset manually.)maxrz3Expect ``mode`` to be one in ["min", "max"]. Found .rwav8ki>wav16kzUnsupported sample rate. Found r rcsg|] }d|dqS)s).0imix_dirr"R/home/ubuntu/vllm_env/lib/python3.10/site-packages/torchaudio/datasets/librimix.py Jsz%LibriMix.__init__..cSsg|]}|jqSr")name)r#pr"r"r'r(Lsz*.wav)rrospathexists RuntimeError ValueErrorrr_TASKS_TO_MIXTUREr&src_dirsrangeglobfilessort)selfrrrrrrr"r%r'__init__,s(  zLibriMix.__init__returncCs||}t|j|d|d}g}t|dD]'\}}t|j||d}|j|jkr:td|jd|d|j||q|j||fS)Nr!rrz"Different waveform shapes. mixed: z, src[z]: ) get_metadatarr enumerateshaper/appendr)r6keymetadatamixedsrcsr$path_srcr"r"r' _load_sampleOs    zLibriMix._load_sampler=cCsZ|j|}tj|j||j}g}|jD]}tj|||j}||q|j||fS)amGet metadata for the n-th sample from the dataset. Args: key (int): The index of the sample to be loaded Returns: Tuple of the following items; int: Sample rate str: Path to mixed audio List of str: List of paths to source audios ) r4r+r,relpathr&rr1r<r)r6r=filename mixed_path srcs_pathsdir_rBr"r"r'r9Zs    zLibriMix.get_metadatacCs t|jS)N)lenr4)r6r"r"r'__len__rs zLibriMix.__len__cCs ||S)a_Load the n-th sample from the dataset. Args: key (int): The index of the sample to be loaded Returns: Tuple of the following items; int: Sample rate Tensor: Mixture waveform List of Tensors: List of source waveforms )rC)r6r=r"r"r' __getitem__us zLibriMix.__getitem__N)rrrr r)__name__ __module__ __qualname____doc__rstrrintr7rtorchTensorrrCr9rJrKr"r"r"r'rs0  # (r)r+pathlibrtypingrrrrRtorch.utils.datartorchaudio.datasets.utilsrr0rr"r"r"r's