Cross-Language Speech Emotion Recognition Using Multimodal Dual Attention Transformers