Skip to main content

Table 10 Average class error and error per class obtained on the Albayzín 2010 test partition for different systems proposed in the literature compared to our proposed RNN approaches

From: Multiclass audio segmentation based on recurrent neural networks for broadcast domain data

SystemClass error (%)Avg
 muspsmsn 
Eval winner [65]19.2039.5025.0037.2030.30
FA HMM [17]18.8023.70*23.6029.1023.80
DCASE Baseline19.0325.5823.5929.5225.18
RNN baseline [52]14.1922.1418.8225.0420.05
RNN + Pool12.8722.5819.1624.9919.90
RNN + Pool + mixup12.4622.8617.3424.3519.25