Rumei Han, Ruining Zhang, Xinhao Yang, Jiaxin Li, Yuan Zhang
Respondent-level data from on-site soundscape surveys (ISO/TS 12913-2 Method A) at 13 urban public spaces in Shenyang and Changshu, China, collected in spring-autumn field campaigns from 2018 to 2025 (2,821 records). Each record combines the questionnaire responses with matched physical measurements: perceived sound sources, the eight ISO 12913 perceptual attributes, ISO pleasantness and eventfulness, overall soundscape evaluation, appropriateness, perceived loudness, current and desired visit frequency, WHO-5 well-being, perceived-restorativeness ratings and participant demographics; per-channel acoustic and psychoacoustic indicators from one-minute binaural recordings made beside each respondent (A-weighted equivalent and percentile levels, level ranges, Zwicker loudness with percentiles and ratios, sharpness, roughness, fluctuation strength and tonality, each with the larger-channel, channel-mean and left-right-difference summaries); and photo-derived semantic-segmentation area shares describing the visual context of each survey point. Data-quality flags mark records whose questionnaire, audio or photograph was excluded during checking. An English variable dictionary and per-site questionnaire, audio and photo coverage notes are included. The data support studies of well-being equity, age differences, soundscape appropriateness, perceived restorativeness and audio-visual interaction in urban public space.