Tanhuasi Park was selected as the case area to collect photos and sounds, and the proportion of each element was identified by image semantic segmentation. Then, according to the saliency and universality of the sound, the sound of birdsong, the sound of water flowing, the sound of wind, and the sound of communication were selected, and the sound and picture were edited by Adobe Premierepro, and the questionnaire was filled in to calculate the recovery value of different sounds. The correlation analysis between landscape elements and sound was carried out, and it was concluded that birdsong had a significant positive correlation with tree and shrub coverage and plant species, water flow sound had a significant positive correlation with water quality and waterscape coverage, and communication sound was negatively correlated with the richness of landscape sketches. Based on this, strategies for improving the green space environment of urban parks are proposed.