Animal sexual signals (e.g., songs, displays) determine who mates with whom and, thus, differences across species and even populations are critical for the build up of reproductive isolation. The evolution of differences across populations has been alternatively argued to be sped up (due to faster evolution of learned traits) or slowed down (due to learning across individuals from different populations) if the signals are socially learned, as is bird song. In this project, our aim is to understand how genetic isolation among populations co-varies with divergence in sexual traits to determine whether variation in learned song can promote reproductive barriers. We have obtained population-level sequencing data from 10 distinct pied flycatcher populations as well as extensive song recordings from each population. I've applied for a continuation from a previous project, since the main questions are the same and we will use very similar resources. I have an additional compute project (NAISS 2024/22-690) at Alvis that utilizes GPU systems for machine learning-based image classification. Given the highly distinct resources and aims of these projects, I apply for a separate compute project here.