Abstract: Recent studies have demonstrated the effectiveness of fine-tuning self-supervised speech representation models for speech emotion recognition (SER). However, applying SER in real-world ...