%0 Conference Paper %1 7952261 %A Gemmeke, Jort F. %A Ellis, Daniel P. W. %A Freedman, Dylan %A Jansen, Aren %A Lawrence, Wade %A Moore, R. Channing %A Plakal, Manoj %A Ritter, Marvin %B 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) %D 2017 %K ma_ss22_ts %P 776-780 %R 10.1109/ICASSP.2017.7952261 %T Audio Set: An ontology and human-labeled dataset for audio events %U https://ieeexplore.ieee.org/document/7952261 %X Audio event recognition, the human-like ability to identify and relate sounds from audio, is a nascent problem in machine perception. Comparable problems such as object detection in images have reaped enormous benefits from comprehensive datasets - principally ImageNet. This paper describes the creation of Audio Set, a large-scale dataset of manually-annotated audio events that endeavors to bridge the gap in data availability between image and audio research. Using a carefully structured hierarchical ontology of 632 audio classes guided by the literature and manual curation, we collect data from human labelers to probe the presence of specific audio classes in 10 second segments of YouTube videos. Segments are proposed for labeling using searches based on metadata, context (e.g., links), and content analysis. The result is a dataset of unprecedented breadth and size that will, we hope, substantially stimulate the development of high-performance audio event recognizers.

@inproceedings{7952261, abstract = {Audio event recognition, the human-like ability to identify and relate sounds from audio, is a nascent problem in machine perception. Comparable problems such as object detection in images have reaped enormous benefits from comprehensive datasets - principally ImageNet. This paper describes the creation of Audio Set, a large-scale dataset of manually-annotated audio events that endeavors to bridge the gap in data availability between image and audio research. Using a carefully structured hierarchical ontology of 632 audio classes guided by the literature and manual curation, we collect data from human labelers to probe the presence of specific audio classes in 10 second segments of YouTube videos. Segments are proposed for labeling using searches based on metadata, context (e.g., links), and content analysis. The result is a dataset of unprecedented breadth and size that will, we hope, substantially stimulate the development of high-performance audio event recognizers.}, added-at = {2022-07-11T22:32:45.000+0200}, author = {Gemmeke, Jort F. and Ellis, Daniel P. W. and Freedman, Dylan and Jansen, Aren and Lawrence, Wade and Moore, R. Channing and Plakal, Manoj and Ritter, Marvin}, biburl = {https://www.bibsonomy.org/bibtex/2d8b7b64976d48e6b7c7d983c05073184/j.haefner}, booktitle = {2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, description = {Audio Set: An ontology and human-labeled dataset for audio events | IEEE Conference Publication | IEEE Xplore}, doi = {10.1109/ICASSP.2017.7952261}, interhash = {b7bd58ee614d3fba2de24df8078198d1}, intrahash = {d8b7b64976d48e6b7c7d983c05073184}, issn = {2379-190X}, keywords = {ma_ss22_ts}, month = {March}, pages = {776-780}, timestamp = {2022-07-11T22:32:45.000+0200}, title = {Audio Set: An ontology and human-labeled dataset for audio events}, url = {https://ieeexplore.ieee.org/document/7952261}, year = 2017 }