Classification on ESC-50

Metric: Accuracy (5-fold) (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide augmentations

Sort:

#	Model↕	Accuracy (5-fold)▼	Augmentations	Paper	Date↕	Code
1	OmniVec2	99.1	Yes	-	-	-
2	InternVideo2	98.6	Yes	InternVideo2: Scaling Foundation Models for Mult...	2024-03-22	Code
3	M2D2 AS+	98.5	Yes	M2D2: Exploring General-purpose Audio-Language R...	2025-03-28	Code
4	OmniVec	98.4	Yes	OmniVec: Learning robust representations with cr...	2023-11-07	-
5	BEATs	98.1	Yes	BEATs: Audio Pre-Training with Acoustic Tokenizers	2022-12-18	Code
6	mn40_as	97.45	Yes	Efficient Large-scale Audio Tagging via Transfor...	2022-11-09	Code
7	DyMN-L	97.4	Yes	Dynamic Convolutional Neural Networks as Efficie...	2023-10-24	Code
8	M2D-CLAP/0.7	97.4	Yes	M2D-CLAP: Masked Modeling Duo Meets CLAP for Lea...	2024-06-04	Code
9	M2D-AS/0.7	97.2	Yes	Masked Modeling Duo: Towards a Universal Audio P...	2024-04-09	Code
10	HTS-AT	97	Yes	HTS-AT: A Hierarchical Token-Semantic Audio Tran...	2022-02-02	Code
11	EAT-M	96.3	Yes	End-to-End Audio Strikes Back: Boosting Augmenta...	2022-04-25	Code
12	ERANN-2-5	96.1	No	-	-	-
13	M2D/0.7	96	Yes	Masked Modeling Duo: Towards a Universal Audio P...	2024-04-09	Code
14	EAT	96	Yes	EAT: Self-Supervised Pre-Training with Efficient...	2024-01-07	Code
15	Audio Spectrogram Transformer	95.7	Yes	AST: Audio Spectrogram Transformer	2021-04-05	Code
16	EAT-S	95.25	Yes	End-to-End Audio Strikes Back: Boosting Augmenta...	2022-04-25	Code
17	MATPAC (SSL model, linear eval)	93.5	No	Masked Latent Prediction and Classification for ...	2025-02-17	Code
18	EAT-S (scratch)	92.15	No	End-to-End Audio Strikes Back: Boosting Augmenta...	2022-04-25	Code
19	SepTr + LeRaC	91.58	No	Learning Rate Curriculum	2022-05-18	Code
20	Multi-Channel Audio Feature with CNN	89.5	No	-	-	-
21	ACDNet	87.1	No	Environmental Sound Classification on the Edge: ...	2021-03-05	Code

#1OmniVec2
99.1
Accuracy (5-fold)· Augmentations
No paper
#2InternVideo2SOTA
98.6
Accuracy (5-fold)· Augmentations· 2024-03-22
InternVideo2: Scaling Foundation Models for Multimodal Video Understanding Code
#3M2D2 AS+
98.5
Accuracy (5-fold)· Augmentations· 2025-03-28
M2D2: Exploring General-purpose Audio-Language Representations Beyond CLAP Code
#4OmniVecSOTA
98.4
Accuracy (5-fold)· Augmentations· 2023-11-07
OmniVec: Learning robust representations with cross modal sharing
#5BEATsSOTA
98.1
Accuracy (5-fold)· Augmentations· 2022-12-18
BEATs: Audio Pre-Training with Acoustic Tokenizers Code
#6mn40_asSOTA
97.45
Accuracy (5-fold)· Augmentations· 2022-11-09
Efficient Large-scale Audio Tagging via Transformer-to-CNN Knowledge Distillation Code
#7DyMN-L
97.4
Accuracy (5-fold)· Augmentations· 2023-10-24
Dynamic Convolutional Neural Networks as Efficient Pre-trained Audio Models Code
#8M2D-CLAP/0.7
97.4
Accuracy (5-fold)· Augmentations· 2024-06-04
M2D-CLAP: Masked Modeling Duo Meets CLAP for Learning General-purpose Audio-Language Representation Code
#9M2D-AS/0.7
97.2
Accuracy (5-fold)· Augmentations· 2024-04-09
Masked Modeling Duo: Towards a Universal Audio Pre-training Framework Code
#10HTS-ATSOTA
97
Accuracy (5-fold)· Augmentations· 2022-02-02
HTS-AT: A Hierarchical Token-Semantic Audio Transformer for Sound Classification and Detection Code
#11EAT-M
96.3
Accuracy (5-fold)· Augmentations· 2022-04-25
End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network Code
#12ERANN-2-5
96.1
Accuracy (5-fold)
No paper
#13M2D/0.7
96
Accuracy (5-fold)· Augmentations· 2024-04-09
Masked Modeling Duo: Towards a Universal Audio Pre-training Framework Code
#14EAT
96
Accuracy (5-fold)· Augmentations· 2024-01-07
EAT: Self-Supervised Pre-Training with Efficient Audio Transformer Code
#15Audio Spectrogram TransformerSOTA
95.7
Accuracy (5-fold)· Augmentations· 2021-04-05
AST: Audio Spectrogram Transformer Code
#16EAT-S
95.25
Accuracy (5-fold)· Augmentations· 2022-04-25
End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network Code
#17MATPAC (SSL model, linear eval)
93.5
Accuracy (5-fold)· 2025-02-17
Masked Latent Prediction and Classification for Self-Supervised Audio Representation Learning Code
#18EAT-S (scratch)
92.15
Accuracy (5-fold)· 2022-04-25
End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network Code
#19SepTr + LeRaC
91.58
Accuracy (5-fold)· 2022-05-18
Learning Rate Curriculum Code
#20Multi-Channel Audio Feature with CNN
89.5
Accuracy (5-fold)
No paper
#21ACDNetSOTA
87.1
Accuracy (5-fold)· 2021-03-05
Environmental Sound Classification on the Edge: A Pipeline for Deep Acoustic Networks on Extremely Resource-Constrained Devices Code