Speech Emotion Recognition using Attention-based LSTM-Network with Residual Connection 事件

Speech Emotion Recognition using Attention-based LSTM-Network with Residual Connection arXiv:2606.03359v1 Announce Type: cross Abstract: Speech emotion recognition is an important component of modern human-computer interaction systems. However, many state-of-the-art approaches rely on large pretrained models with high computational and memory requirements, limiting their applicability. This paper proposes ResLSTM-SA, a lightweight architecture that integrates residual connections with soft atte