Fig. 4From: AUC optimization for deep learning-based voice activity detectionROC curves of the comparison methods with the BLSTM model and MRCG feature on the Chinese Noisy-THCHS-30 test dataset at \(-5\) dBBack to article page