Disentangled representation learning

The article introduces DSNet, a Disentangled Siamese Network with neutral calibration for speech emotion recognition (SER). DSNet aims to address the challenges in SER, such as the unconscious encoding of…