aac_datasets.datasets.functional.wavcaps module¶

class WavCapsCard[source]¶

Bases: DatasetCard

ANNOTATIONS_CREATORS: Tuple[str, ...] = ('machine-generated',)¶

CAPTIONS_PER_AUDIO: Dict[str, int] = {'audioset': 1, 'audioset_no_audiocaps': 1, 'bbc': 1, 'freesound': 1, 'freesound_no_clotho': 1, 'freesound_no_clotho_v2': 1, 'soundbible': 1}¶

CITATION: str = '\n @article{mei2023WavCaps,\n title = {Wav{C}aps: A {ChatGPT}-Assisted Weakly-Labelled Audio Captioning Dataset for Audio-Language Multimodal Research},\n author = {Xinhao Mei and Chutong Meng and Haohe Liu and Qiuqiang Kong and Tom Ko and Chengqi Zhao and Mark D. Plumbley and Yuexian Zou and Wenwu Wang},\n year = 2023,\n journal = {arXiv preprint arXiv:2303.17395},\n url = {https://arxiv.org/pdf/2303.17395.pdf}\n }\n '¶

DEFAULT_REVISION: str = '85a0c21e26fa7696a5a74ce54fada99a9b43c6de'¶

DEFAULT_SUBSET: str = 'audioset_no_audiocaps'¶

DESCRIPTION: str = 'WavCaps: A ChatGPT-Assisted Weakly-Labelled Audio Captioning Dataset for Audio-Language Multimodal Research.'¶

EXPECTED_SIZES: Dict[str, int] = {'AudioSet_SL': 108317, 'BBC_Sound_Effects': 31201, 'FreeSound': 262300, 'SoundBible': 1320}¶

HOMEPAGE = 'https://huggingface.co/datasets/cvssp/WavCaps'¶

LANGUAGE: Tuple[str, ...] = ('en',)¶

LANGUAGE_DETAILS: Tuple[str, ...] = ('en-US',)¶

NAME: str = 'wavcaps'¶

PRETTY_NAME: str = 'WavCaps'¶

REPO_ID: str = 'cvssp/WavCaps'¶

SAMPLE_RATE: int = 32000¶

SIZE_CATEGORIES: Tuple[str, ...] = ('100K<n<1M',)¶

SOURCES: Tuple[str, ...] = ('AudioSet_SL', 'BBC_Sound_Effects', 'FreeSound', 'SoundBible')¶

SUBSETS: Tuple[str, ...] = ('audioset', 'bbc', 'freesound', 'soundbible', 'audioset_no_audiocaps', 'freesound_no_clotho', 'freesound_no_clotho_v2')¶

TASK_CATEGORIES: Tuple[str, ...] = ('audio-to-text', 'text-to-audio')¶

download_wavcaps_dataset( root: str | Path | None = None, subset: str = 'audioset_no_audiocaps', force: bool = False, verbose: int = 0, verify_files: bool = False, clean_archives: bool = False, hf_cache_dir: str | None = None, repo_id: str | None = None, revision: str | None = None, zip_path: str | Path | None = None, ) → None[source]¶

Prepare WavCaps data.

Parameters:

root – Dataset root directory. defaults to “.”.
subset – The subset of MACS to use. Can be one of SUBSETS. defaults to “audioset_no_audiocaps”.
force – If True, force to download again all files. defaults to False.
verbose – Verbose level. defaults to 0.
verify_files – If True, check all file already downloaded are valid. defaults to False.
clean_archives – If True, remove the compressed archives from disk to save space. defaults to True.
hf_cache_dir – Optional override for HuggingFace cache directory path. defaults to None.
repo_id – Repository ID on HuggingFace. defaults to “cvssp/WavCaps”.
revision – Optional override for revision commit/name for HuggingFace rapository. defaults to None.
zip_path – Path to zip executable path in shell. defaults to “zip”.

download_wavcaps_datasets( root: str | Path | None = None, subsets: str | Iterable[str] = 'audioset_no_audiocaps', force: bool = False, verbose: int = 0, clean_archives: bool = False, hf_cache_dir: str | None = None, repo_id: str | None = None, revision: str | None = None, verify_files: bool = False, zip_path: str | Path | None = None, ) → None[source]¶: Function helper to download a list of subsets. See download_wavcaps_dataset() for details.

load_wavcaps_dataset( root: str | Path | None = None, subset: str = 'audioset_no_audiocaps', verbose: int = 0, hf_cache_dir: str | None = None, revision: str | None = None, ) → Dict[str, List[Any]][source]¶

Load WavCaps metadata.

Parameters:

root – Dataset root directory. defaults to “.”.
subset – The subset of MACS to use. Can be one of SUBSETS. defaults to “audioset_no_audiocaps”.
verbose – Verbose level. defaults to 0.
hf_cache_dir – Optional override for HuggingFace cache directory path. defaults to None.
revision – Optional override for revision commit/name for HuggingFace rapository. defaults to None.

Returns:

A dictionnary of lists containing each metadata.