ResonaVis：可视化交互式音乐数据以支持治疗场景下的反思性音乐创作
ResonaVis: Visualizing Interactive Music Data to Support Reflective Music Composition for Therapeutic Contexts
浏览论文内容
中文总结 AI 辅助
ResonaVis是交互式可视化系统，整合音频与互动数据为治疗场景音乐创作提供数据支持，经评估可用性良好，可提升创作者的分析推理与创作决策信心。
中文摘要 AI 辅助
为治疗场景创作音乐需要在音乐结构与听众感官反应之间协调复杂关系，然而作曲家往往缺乏这些互动的结构化表征，只能依赖直觉。我们提出ResonaVis，这是一个交互式可视化系统，可帮助作曲家分析之前与自闭症谱系障碍（ASD）儿童开展的会话中的互动数据和音频数据，为未来的创作提供参考。ResonaVis整合了音频特征和互动日志，以捕捉儿童如何参与分层音乐创作，并通过时间转换、层共现、节奏活动和频谱特征的协同可视化来呈现这种参与情况。该系统并不直接规定策略或支持治疗会话，而是呈现过往会话数据中的模式，以支持创作过程中基于数据的反思。我们通过一项包含8名音乐学生的混合方法研究和一项针对2名经验丰富作曲家的后续案例研究对ResonaVis进行评估。结果显示其可用性良好（系统可用性量表SUS=72.23），超过早期原型的基准，且参与者能够识别互动模式、推理层关系，并以高感知性能和低挫败感做出合理的创作决策。在多个维度上，解读ASD相关创作的互动与声学数据的信心显著提升（p<0.05），定性发现表明创作转向更具适应性、基于数据的模式。本研究贡献了治疗性音乐互动数据的可视化设计空间、用于创作反思的集成系统，以及可视化工具支持分析推理和基于数据的创意实践信心的实证证据。
英文摘要
Designing music for therapeutic contexts requires navigating complex relationships between musical structure and listeners' sensory responses, yet composers often lack structured representations of these interactions, relying instead on intuition. We present ResonaVis, an interactive visualization system that helps composers analyze interaction and audio data from prior sessions with children with Autism Spectrum Disorder (ASD), informing future compositions. ResonaVis integrates audio features and interaction logs to capture how children engage with layered musical compositions, representing this engagement through coordinated visualizations of temporal transitions, layer co-occurrence, rhythmic activity, and spectral characteristics. Rather than prescribing strategies or supporting therapy sessions directly, the system surfaces patterns in past session data to support data-informed reflection during composition. We evaluated ResonaVis through a mixed-methods study with eight music students and a follow-up case study with two experienced composers. Results demonstrate good usability (SUS = 72.23), exceeding benchmarks for early prototypes, and show that participants could identify interaction patterns, reason about layer relationships, and make informed compositional decisions with high perceived performance and low frustration. Confidence in interpreting interaction and acoustic data for ASD-focused composition increased significantly across multiple dimensions (p < 0.05), with qualitative findings suggesting a shift toward more adaptive, data-informed composition. This work contributes a visualization design space for therapeutic music interaction data, an integrated system for compositional reflection, and empirical evidence that visualization tools support analytical reasoning and confidence in data-informed creative practice. (Abstract shortened for arXiv.)
