Abstract: Chord recognition as the basis of automatic musical information label plays an important role in analyzing the music structure, transcribing the note and identifying the melody. According to the related music theory this paper proposes a new chord recognition approach based on the beat synchronization and the SRC（Sparse Representation-
based Classification）. Different from the traditional frame-based methods, the proposed method considers the beat as the fundamental time slot for chord variation. It applies CQT（Constant-Q Transform） to perform the time-frequency analysis to the music signal to abstract the PCP（Pitch Class Profile） feature, and recognizes the chords by SRC. The experimental results show that the new method is better than the template-based method.

Key words: chord recognition, beat tracking, Pitch Class Profile（PCP）, Sparse Representation-based Classification（SRC）

摘要： 和弦识别作为音乐信息标注的基础，在分析音乐结构和旋律方面具有非常重要的作用。结合音乐理论知识，提出一种基于稀疏表示分类器的和弦识别方法。与传统的基于帧的识别方法不同，以节拍作为和弦变化的最小时间间隔，利用CQT（Constant-Q Transform）变换对音乐信号进行时频分析，提取PCP（Pitch Class Profile）特征，采用稀疏表示分类器（Sparse Representation-based Classification，SRC）进行和弦识别。实验结果表明，提出的特征和识别方法在识别率上均高于传统的方法。

关键词: 和弦识别, 节拍跟踪, 音级轮廓（PCP）, 稀疏表示分类器