EURASIP Journal on Audio, Speech, and Music Processing

Table 1 The average localization performance of the DNN trained with different training set sizes and nodes number in the hidden layer

From: Binaural sound localization based on deep neural network and affinity propagation clustering in mismatched HRTF condition

	200 nodes	500 nodes	1000 nodes
25,000 sounds	42.08%	42.15%	42.41%
2500 sounds	41.32%	41.39%	42.23%

Back to article page