Table A & table B.

Urban acoustic environments significantly influence quality of life, prompting a paradigm shift from noise reduction to soundscape design that considers human perception. This study investigates urban soundscape perceptions using social media data to understand how acoustic environments reflect urban spatial characteristics. We collected Korean-language Twitter texts from 156 points of interest in Seoul’s central and southeastern neighborhoods from October to November 2020. The collected texts were preprocessed through text cleaning and morphological analysis, then classified using an urban soundscape taxonomy based on Canter’s place model that integrates physical setting, activity, and meaning dimensions. The results revealed that activity-related soundscapes (79.4%) dominated all target areas, and strong negative correlations between indoor, mechanical, and behavioral soundscapes suggest a spatial separation of acoustic zones by urban function. Especially, historical central districts (22.0%) showed a higher proportion of meaning-related soundscapes compared to modern commercial districts (13.4%). K-means clustering further identified three soundscape types: infrastructure-dominated, socially vibrant, and commercial indoor-focused zones. By combining social media analysis with spatial patterns, the study provides a scalable tool for urban planners to understand how citizens experience urban soundscapes, offering data-driven insights for contextualized urban design.城市声环境对生活质量具有显著影响，推动了研究范式从噪声治理转向关注人类感知的声景（soundscape）设计。本研究借助社交媒体数据探究城市声景感知，以解析声环境如何反映城市空间特征。研究于2020年10月至11月期间，收集了首尔市中心及东南部片区156个兴趣点（Point of Interest, POI）的韩语Twitter文本数据。所收集的文本经文本清洗与形态分析预处理后，基于整合了物理环境、活动与意义三个维度的坎特场所模型（Canter's place model）构建的城市声景分类体系完成分类。研究结果显示，与活动相关的声景占比达79.4%，在所有目标区域中占据主导地位；室内声景、机械声景与行为声景间存在显著负相关，表明声学区划随城市功能呈现空间分离特征。尤为值得关注的是，历史中心城区的意义相关声景占比达22.0%，较现代商业区的13.4%更高。K均值聚类（K-means clustering）进一步识别出三类声景类型：以基础设施为主导型、社交活跃型以及聚焦商业室内型区域。本研究将社交媒体分析与空间格局分析相结合，为城市规划者提供了可扩展的分析工具，用以理解市民对城市声景的体验感知，并为情境化城市设计提供数据驱动的决策参考。