Multi-Loss Learning for Speech Emotion Recognition with Energy-Adaptive Mixup and Frame-Level Attention
多损失学习用于带有能量自适应混排和帧级注意力的语音情感识别
机构 * Beijing University of Posts and Telecommunications(北京邮电大学) ; Li Auto(利汽车)
AI总结 本文提出多损失学习框架,结合能量自适应混排和帧级注意力模块,以提升语音情感识别的性能和鲁棒性。
Comments Submitted for review to Interspeech 2026