@inproceedings{bellver-soler-etal-2025-cutting,
    title = "Cutting Through Overload: Efficient Token Dropping for Speech Emotion Recognition in Multimodal Large Language Models",
    author = "Bellver-Soler, Jaime  and
      Rodriguez-Cantelar, Mario  and
      C{\'o}rdoba, Ricardo  and
      D{'}Haro, Luis Fernando",
    editor = "Torres, Maria Ines  and
      Matsuda, Yuki  and
      Callejas, Zoraida  and
      del Pozo, Arantza  and
      D'Haro, Luis Fernando",
    booktitle = "Proceedings of the 15th International Workshop on Spoken Dialogue Systems Technology",
    month = may,
    year = "2025",
    address = "Bilbao, Spain",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/fix-sig-urls/2025.iwsds-1.30/",
    pages = "284--289",
    ISBN = "979-8-89176-248-0",
    abstract = "Recent developments in Multimodal Large Language Models (MLLMs) have provided novel insights into Speech Emotion Recognition (SER). However, combining high-dimensional speech signals with textual tokens can lead to a rapid growth in input tokens, increasing computational costs and inference times. This ``token overload'' also risks shadowing essential textual cues, affecting the reasoning capabilities of the language model and diluting emotional information crucial to accurate SER. In this paper, we explore different token drop methods that mitigate excessive token counts while preserving both emotional nuances and the core linguistic capabilities of the model. Specifically, we compare various pooling approaches to produce a compact representation. Our preliminary findings suggest that these techniques can reduce computational costs without decreasing SER accuracy."
}