ESC: Dataset for Environmental Sound Classification

By Karol J. Piczak. 2015

2,068 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Developmental Biology, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Developmental Biology

In Signal Processing

Citations in the corpus

Self-Supervised MultiModal Versatile Networks (2020, methodology): “Some classification datasets have official splits (3 for UCF101/HMDB51 and 5 for ESC-50).”
BigSSL: Exploring the Frontier of Large-Scale Semi-Supervised Learning for Automatic Speech Recognition (2021, background): “mask detection during speech [90], environmental human sound detection such as coughing and sneezing [91] and dementia detection [92].”
Look, Listen and Learn (2017, positive): “For 5 second recordings from ESC-50 we extract 10 equally spaced 1 second subclips, while for the 6 times longer DCASE recordings, 60 subclips are extracted per clip.”
SoundNet: Learning Sound Representations from Unlabeled Video (2016, methodology): “In both evaluations a multi-class SVM (multiple one-vs all classifiers) is trained over extracted Accuracy on Comparison of SoundNet Model ESC-50 ESC-10 Loss 8 Layer, `2 Loss 47.8% 81.5%8 Layer, KL Loss 72.9% 92.2% Teacher Net 8 Layer, ImageNet Only 69.5% 89.8% 8 Layer, Places Only 71.1% 89.5% 8 Lay”
FSD50K: An Open Dataset of Human-Labeled Sound Events (2020, methodology): “Before the release of AudioSet, the most widely used datasets for SET have been UrbanSound8K [39], ESC-50 [22], and to a lesser extent CHiME-home [21].”
VATT: Transformers for Multimodal Self-Supervised Learning from Raw Video, Audio and Text (2021, methodology): “We use ESC50 to train a linear classifier on top of the frozen audio Transformer.”
Cooperative Learning of Audio and Video Models from Self-Supervised Synchronization (2018, methodology): “For this purpose, we take the activations from the conv_5 layer of the audio subnetwork and test their quality as audio representation on two established sound classiﬁcation datasets: ESC-50 [29] and DCASE2014 [30].”
VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning (2021, positive): “We demonstrate the ability of VICReg to function in a setting where the branches have different architectures by pretraining on the ESC-50 audio dataset Piczak (2015), which is an environmental sound classification dataset with 50 classes.”
Audio-Visual Instance Discrimination with Cross-Modal Agreement (2020, methodology): “– Audio Features: We evaluate the audio features on the ESC-50 [62] dataset by training linear classiﬁers on ﬁxed features from the pre-trained audio network.”
PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition (2019, background): “1) ESC-50: ESC-50 is an environmental sound dataset [50] consisting of 50 sound events, such as “Dog” and “Rain”.”
Addressing Missing Labels in Large-Scale Sound Event Recognition Using a Teacher-Student Framework With Loss Masking (2020, background): “As Sound Event Recognition (SER) has gained increasing attention in recent years [1], research in this field has moved from small datasets encompassing few hours of audio [2, 3, 4, 5], to larger datasets with much greater coverage and duration [6, 7].”
Environmental sound classification with convolutional neural networks (2015, methodology): “5), when compared with its analogue for the unsupervised technique presented by Salamon & Bello [37], shows some complementary characteristics.”
Balanced Multimodal Learning via On-the-fly Gradient Modulation (2022, methodology): “We take L(3)-Net framework [1], Audio-visual Scene Analysis [3], and Audio-visual Coattention [2], respectively, to perform multimodal selfsupervised pretraining on Kinetics-Sounds and then conduct audio classification with the pretrained audio encoder on ESC50 [4] to evaluate the representation qua”
AudioCaps: Generating Captions for Audios in The Wild (2019, background): “There exist a number of datasets to aid in achieving this goal, including DCASE series (Stowell et al., 2015; Mesaros et al., 2016, 2017), UrbanSound8k (Salamon et al., 2014), ESC (Piczak, 2015).”
SSAST: Self-Supervised Audio Spectrogram Transformer (2021, methodology): “ESC-50 (ESC) We use the ESC-50 dataset (Piczak 2015) for the single-label audio event classification task.”
Audio-Visual Model Distillation Using Acoustic Images (2019, background): “Environmental Sound Classification (ESC-50) [36] is a labeled collection of 2.000 environmental audio recordings manually extracted from Freesound.”
BEATs: Audio Pre-Training with Acoustic Tokenizers (2022, background): “BEAT S achieve superior performance across six audio and speech classiﬁcation tasks and set new state-of-the-art results on AudioSet-2M and ESC-50 benchmarks.”