基于视觉缩略图的高维音乐信息可视化的关键技术研究

Visualizing multivariate musical content based on visual thumbnails

对基于视觉缩略图的高维音乐信息可视化技术作了深入探讨。对代表性的音乐可视化、特别是视觉缩略图的各种技术进行了广泛调研,通过一系列的用户调查分析了音乐内容的视觉缩略图应具备的基本特征。在此基础之上,提出了一种新颖的视觉缩略图ThumbnailDJ,并对其进行了一系列的用户测试。在对实验结果进行分析后,讨论了音乐信息可视化的应用前景、发展方向及相关研究重点。高维音乐内容的视觉描述有助于提高音乐库的浏览和检索效率,本研究将有助于缩小音乐视觉描述与用户音乐感知之间的语义鸿沟,提高音乐库的浏览和检索效率。同时,研究成果也将对高维数据的信息可视化研究起到重要的借鉴作用。

This paper presented an exploration of visualizing multivariate musical content based on visual thumbnails.Firstly it explored representative applications of musical content visualization.Based on a series of user investigations, analyzed the fundamental features of visual thumbnails for musical content.Then developed a novel concept called ThumbnailDJ,and conducted several rounds of user studies.Based on the accumulated insights,discussed the development directions of this topic and suggested relevant key research issues.Visualization of multivariate musical content facilitates browsing and retrieval in music collections.This work can substantially minimize the semantic gap between visual thumbnails and human musical perception,hence improve the efficiency of browsing and retrieval in music databases.Meanwhile,relevant research outcomes will provide valuable insights for the visualization of multivariate data in general.