Identifying "Cover Songs" with Beat-Synchronous Chroma Features
Identifying "Cover Songs" with Beat-Synchronous Chroma Features
复制标题
DOI：
10.7916/d8b284n1
复制
发表时间：
2006
期刊：
影响因子：
--
通讯作者：
D. Ellis;Graham E. Poliner
中科院分区：
文献类型：
--
作者：
D. Ellis;Graham E. Poliner
大型音乐收藏（从数千首到数百万首曲目）不适合手动搜索，这激发了自动搜索方法的开发。当两个音乐团体表演相同的歌曲或作品时，这些被称为“翻唱”版本。我们描述了一种试图识别音乐音频记录之间的这种关系的系统。为了克服克里思的可变性，我们使用节拍跟踪来描述每个节拍的特征向量。为了处理仪器中的变化，我们使用12维色度特征向量，该特征向量收集支持八度音阶的每个八度音阶的频谱能量。为了比较两个记录，我们简单地将两个轨道的整个逐色度差表示进行交叉相关，并寻找指示片段之间良好局部对准的尖峰。在USPOP2002集合中识别的15对流行音乐封面版本的小集合上的评估实现了约60%正确的性能。
Large music collections, ranging from thousands to millions of tracks, are unsuited to manual searching, motivating the development of automatic search methods. When two musical groups perform the same underlying song or piece, these are known as ‘cover’ versions. We describe a system that attempts to identify such a relationship between music audio recordings. To overcome variability in tempo, we use beat-tracking to describe each piece with one feature vector per beat. To deal with variation in instrumentation, we use 12-dimensional chroma feature vectors that collect spectral energy supporting each semitone of the octave. To compare two recordings, we simply cross-correlate the entire beat-by-chroma representation for two tracks and look for sharp peaks indicating good local alignment between the pieces. Evaluation on a small set of 15 pairs of pop music cover versions identified within the USPOP2002 collection achieves a performance of around 60% correct.