EURASIP Journal on Audio, Speech, and Music Processing

Table 8 Performance (in overall SAD detection error [77]) of deep learning-based approaches vs. the proposed algorithm for room-localized SAD on the DIRHA-sim-evalita corpus

From: Room-localized speech activity detection in multi-microphone smart homes

Method		SAD error(%)
Deep learning	DNN [30]	5.8
	3D-CNN [31]	7.0
	3D-CNN [32]	5.2
	3D-CNN (SAD+SLOC) [32]	3.5
Proposed	Seg	5.7
	Win	4.7

Back to article page