Collins Nick
Music Department, Durham University, Durham, UK.
R Soc Open Sci. 2025 Jul 9;12(7):241647. doi: 10.1098/rsos.241647. eCollection 2025 Jul.
Audio content analysis can be deployed to examine relationships within and between collected works of different music artists, allowing a new approach to comparative analysis of recorded music within the domain of computational musicology. Although current-generation automatic transcription retains some flaws with respect to expert human analysis, there is a consistency to applying the same algorithms on disparate works, and the benefit of tireless calculation with explicit open bias. In the present study, three successful alternative rock groups, and three 'control' artists, all from either the United States or the UK, are compared with respect to their musical careers through their main recorded releases (spanning the years 1983-2021 for the main three and 1957-2000 for the controls). Statistical measures of variation over time, and the diversity of their recorded output, are used to answer research questions on their studio career and the originality of their work. The techniques explored here are immediately pertinent to study other artists outside of this starting point, and we discuss the potential and challenges of such approaches for the musicology of recorded music.
音频内容分析可用于研究不同音乐艺术家的作品集内部及之间的关系，为计算音乐学领域内录制音乐的比较分析提供一种新方法。尽管当前一代的自动转录与专家人工分析相比仍存在一些缺陷，但在不同作品上应用相同算法具有一致性，且有无休止计算和明确公开偏差的优势。在本研究中，对来自美国或英国的三个成功的另类摇滚乐队和三位“对照”艺术家的音乐生涯进行了比较，比较依据是他们的主要录音作品发行情况（三个主要乐队的时间跨度为1983年至2021年，对照艺术家的时间跨度为1957年至2000年）。通过对随时间变化的统计度量以及他们录制作品的多样性，来回答关于他们录音生涯和作品原创性的研究问题。这里探索的技术对于研究这一起始范围之外的其他艺术家具有直接相关性，并且我们讨论了此类方法在录制音乐音乐学方面的潜力和挑战。
