Sound Event Detection2020en

Domestic Environment Sound Event Detection (DCASE Task 4)

Sound event detection dataset covering 10 domestic sound classes. Used as the benchmark for DCASE Challenge Task 4. Primary metric is event-based macro-F1 on the public eval / validation set.

Current State of the Art

ATST-SED

Audio Research

58.1

event-f1

DESED — event-f1

3 results · 1 SOTA advances · higher is better

All results
SOTA frontier
404550556020262027event-f1ATST-SED

event-f1 Progress Over Time

Showing 2 breakthroughs from Mar 2022 to Apr 2024

43.847.751.655.559.4Mar 2022Apr 2024event-f1Date

Key Milestones

Mar 2022
FDY-CRNN

FDY-CRNN, DESED validation, event-based F1. Nam et al. ICASSP 2022.

45.1
Apr 2024
ATST-SEDCurrent SOTA

ATST-SED, DESED public eval, event-based F1. seed — verify.

58.1
+28.8%
Total Improvement
28.8%
Time Span
2y 1m
Breakthroughs
2
Current SOTA
58.1

Top Models Performance Comparison

Top 3 models ranked by event-f1

event-f11ATST-SED58.1100.0%2FDY-CRNN45.177.6%3Prototypical CRNN (DCASE2...40.169.0%0%25%50%75%100%% of best
Best Score
58.1
Top Model
ATST-SED
Models Compared
3
Score Range
18.0

event-f1Primary

#ModelScorePaper / CodeDate
1
ATST-SEDOpen Source
Audio Research
58.1Apr 2026
2
FDY-CRNNOpen Source
KAIST
45.1Apr 2026
3
Prototypical CRNN (DCASE2021 baseline)Open Source
DCASE
40.1Apr 2026