AudioReranking¶

Number of tasks: 5

ESC50AudioReranking¶

ESC-50 environmental sound dataset adapted for audio reranking. Given a query audio of environmental sounds, rank 5 relevant audio samples higher than 16 irrelevant ones from different sound classes. Contains 200 queries across 50 environmental sound categories for robust evaluation.

Dataset: mteb/ESC50AudioReranking • License: cc-by-3.0 • Learn more →

Task category	Score	Languages	Domains	Annotations Creators	Sample Creation
audio to audio (a2a)	map_at_1000	zxx	AudioScene	expert-annotated	found

Citation

@inproceedings{piczak2015dataset,
  author = {Piczak, Karol J.},
  booktitle = {Proceedings of the 23rd {Annual ACM Conference} on {Multimedia}},
  date = {2015-10-13},
  doi = {10.1145/2733373.2806390},
  isbn = {978-1-4503-3459-4},
  location = {{Brisbane, Australia}},
  pages = {1015--1018},
  publisher = {{ACM Press}},
  title = {{ESC}: {Dataset} for {Environmental Sound Classification}},
  url = {http://dl.acm.org/citation.cfm?doid=2733373.2806390},
}

FSDnoisy18kAudioReranking¶

FSDnoisy18k sound event dataset adapted for audio reranking. Given a query audio with potential label noise, rank 4 relevant audio samples higher than 16 irrelevant ones from different sound classes. Contains 200 queries across 20 sound event categories.

Dataset: mteb/FSDnoisy18kAudioReranking • License: cc-by-4.0 • Learn more →

Task category	Score	Languages	Domains	Annotations Creators	Sample Creation
audio to audio (a2a)	map_at_1000	eng	AudioScene	human-annotated	found

Citation

@inproceedings{fonseca2019fsdnoisy18k,
  author = {Fonseca, Eduardo and Plakal, Manoj and Ellis, Daniel P. W. and Font, Frederic and Favory, Xavier and Serra, Xavier},
  booktitle = {ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  organization = {IEEE},
  pages = {21--25},
  title = {Learning Sound Event Classifiers from Web Audio with Noisy Labels},
  year = {2019},
}

GTZANAudioReranking¶

GTZAN music genre dataset adapted for audio reranking. Given a query audio from one of 10 music genres, rank 3 relevant audio samples higher than 10 irrelevant ones from different genres. Contains 100 queries across 10 music genres for comprehensive evaluation.

Dataset: mteb/GTZANAudioReranking • License: not specified • Learn more →

Task category	Score	Languages	Domains	Annotations Creators	Sample Creation
audio to audio (a2a)	map_at_1000	zxx	Music	human-annotated	found

Citation

@article{1021072,
  author = {Tzanetakis, G. and Cook, P.},
  doi = {10.1109/TSA.2002.800560},
  journal = {IEEE Transactions on Speech and Audio Processing},
  keywords = {Humans;Music information retrieval;Instruments;Computer science;Multiple signal classification;Signal analysis;Pattern recognition;Feature extraction;Wavelet analysis;Cultural differences},
  number = {5},
  pages = {293-302},
  title = {Musical genre classification of audio signals},
  volume = {10},
  year = {2002},
}

UrbanSound8KAudioReranking¶

UrbanSound8K urban sound dataset adapted for audio reranking. Given a query audio of urban sounds, rank 4 relevant audio samples higher than 16 irrelevant ones from different urban sound classes. Contains 200 queries across 10 urban sound categories for comprehensive evaluation.

Dataset: mteb/UrbanSound8KAudioReranking • License: cc-by-4.0 • Learn more →

Task category	Score	Languages	Domains	Annotations Creators	Sample Creation
audio to audio (a2a)	map_at_1000	zxx	Spoken	human-annotated	found

Citation

@inproceedings{Salamon:UrbanSound:ACMMM:14,
  author = {Salamon, Justin and Jacoby, Christopher and Bello, Juan Pablo},
  booktitle = {Proceedings of the 22nd ACM international conference on Multimedia},
  organization = {ACM},
  pages = {1041--1044},
  title = {A Dataset and Taxonomy for Urban Sound Research},
  year = {2014},
}

VocalSoundAudioReranking¶

VocalSound dataset adapted for audio reranking. Given a query vocal sound from one of 6 categories, rank 4 relevant vocal samples higher than 16 irrelevant ones from different vocal sound types. Contains 198 queries across 6 vocal sound categories for robust evaluation.

Dataset: mteb/VocalSoundAudioReranking • License: cc-by-sa-4.0 • Learn more →

Task category	Score	Languages	Domains	Annotations Creators	Sample Creation
audio to audio (a2a)	map_at_1000	eng	Spoken	human-annotated	found

Citation

@inproceedings{Gong_2022,
  author = {Gong, Yuan and Yu, Jin and Glass, James},
  booktitle = {ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  doi = {10.1109/icassp43922.2022.9746828},
  month = may,
  publisher = {IEEE},
  title = {Vocalsound: A Dataset for Improving Human Vocal Sounds Recognition},
  url = {http://dx.doi.org/10.1109/ICASSP43922.2022.9746828},
  year = {2022},
}