Supporting Data for “Agent-Based Models of Sexual Selection in Bird Vocalizations Using Generative Approaches”

Associated publication:Zhao, H.; Arita, T.; Suzuki, R. Agent-Based Models of Sexual Selection in Bird Vocalizations Using Generative Approaches. Appl. Sci. 2025, 15, 10481. https://doi.org/10.3390/app151910481Description:This repository contains the datasets and code associated with the above publication.It includes:The original field recording dataThe training codes and dataset of VAE modelThe trained model checkpoint files (network files) of VAE modelThe codes and evolutionary results for implementing the VAE-based evolutionary modelThe scripts for generating the main results of VAE evolutionary modelThe training dataset of TTA modelThe codes and evolutionary results for implementing the TTA-based evolutionary modelThe scripts for generating the main results of TTA evolutionary modelContents:original_recording_data.zip – Field recordings of the Blue-and-white Flycatcher (Cyanoptila cyanomelana) used as source material.vae_Dataset.zip – Dataset used for VAE model trainingconvolutional-variational-autoencoder.ipynb – VAE training notebookvae_4sec_epoch_2000.pth - The trained VAE model checkpoint files (network files)vae_based_evolutionary_model.ipynb – Implementation of the agent-based evolutionary model using VAE and the analysis of evolutionaryVAE_evo_results.zip - The evolutionary results of VAE-based evolutionary experimentsgeneratedist.ipynb and getImages.ipynb: Helper scripts for visualizing analysis of VAE evolved genes and VAE latent spaceTTA_dataset.rar -– Dataset used for TTA model trainingTTA_based_evolution_model.ipynb – Implementation of the agent-based evolutionary model using fine-tuned TTA modeltta_evo_results.zip - The evolutionary results of TTA-based evolutionary experimentsAnalysis_tta.zip - Helper scripts for generating the main results of TTA- evolutionary modelNotes:Other materials related to the Text-to-Audio (TTA) evolutionary model are stored in a separate folder due to Stability AI Community License requirements. These materials include:TTA model training codesTrained model checkpoint files (network files)All above materials related to TTA model training are available at: https://nuss.nagoya-u.ac.jp/s/FnATf3dJBfTEBTg. This folder also contains detailed documentation on file descriptions and licensing terms.