EURASIP Journal on Audio, Speech, and Music Processing

Table 1 The SELD performance of the proposed method and comparison models for FOA dataset

From: Dual-branch attention module-based network with parameter sharing for joint sound event detection and localization

Models	ER_20°↓	F_20°↑	LE_CD↓	LR_CD↑	SELD↓
DCASE2020task3	0.580	51.3%	18.3°	69.9%	0.367
SELDnet	0.720	37.4%	22.8°	60.7%	0.466
Two-stage network	0.399	67.5%	14.8°	73.8%	0.267
Cao_Surrey	0.363	71.2%	13.3°	81.1%	0.229
Nguyen_NTU	0.360	71.9%	12.1°	82.7%	0.220
Proposed method	0.347	74.0%	9.30°	80.2%	0.214

The “↓” after the evaluation metric in the table indicates that the lower the metric is better, and “↑” indicates that the higher the metric is better

Back to article page