From September 1, 2023 to November 6, 2023
UNIZAR and CONICET are collaborating to advance automated systems for diagnosing, treating, and monitoring voice disorders. The goal is to assist doctors in providing timely care to patients and efficiently identifying those in need of in-person medical attention. This task is highly challenging due to the limited availability of speech data for analysis.
Cyntia Bonomi and Mariel Esteve from CONICET will be working in Zaragoza alongside researchers from VivoLab from September 1st to November 6th, 2023. Specifically, Cyntia will explore data representation issues and the optimal utilization of pretrained self-supervised models for this task. Meanwhile, Mariel will investigate the significance of domain mismatch and other variability factors apart from the presence of the voice disorder. Both researchers will address the challenge of data scarcity in this context and its impact on system performance. They will explore various approaches for model training, adaptation, and calibration to enhance the system's overall performance.
