Anonymized Urban Soundscape Prediction Dataset and Reproducible Code for Street-View and Audio Analysis

This dataset contains anonymized processed data and reproducible code supporting a study on urban soundscape prediction using street-view visual features and audio-derived soundscape labels. The release includes the final modeling dataset, ADE20K visual feature statistics, Qwen3-Omni soundscape classification outputs, a human validation subset, data documentation, and reproducible Python scripts. Raw videos, raw audio files, original street-view images, exact GPS coordinates, local file paths, and original file names are not included for privacy and copyright reasons.