LGFN: A Dynamic Gating Framework for Lyrics-Audio Alignment in Music Emotion Recognition
Yinghui Wang, Mengqi Zang, Huasen Zhang, Xin Shan, Xinyue Wang
This paper introduces LGFN, a cross-modal dynamic fusion framework for music emotion recognition that improves accuracy by aligning lyrics and audio features through attentional alignment and dynamic gate fusion.
bioRxiv · Aug 11, 2025 · PreprintBioengineering