EURASIP Journal on Audio, Speech, and Music Processing

Table 4 The SELD performance of the proposed method and without DBAM method

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↓
Without DBAM	0.416	68.7%	12.492°	74.7%	0.263
DBAM 1	0.367	71.0%	11.248°	79.6%	0.230
DBAM 2	0.347	74.0%	10.818°	80.2%	0.216
DBAM 4	0.376	71.5%	9.304°	80.6%	0.227

Back to article page