特征提取是音乐信号分析中的重要步骤，本文全面总结了音乐信号分析中的音频特征，分为传统音频特征、音乐相关特征和面向深度学习的音频特征三个部分进行介绍，重点梳理了各类主流音频特征的含义、计算方法及音乐信号特征在传统机器学习方法和深度学习方法下的应用现状，并总结了特征提取常用的主流开源工具及其特点。最后，文章对音乐特征领域未来的发展方向进行了展望。

Feature extraction is an important step in music signal analysis. This paper comprehensively summarizes the audio features in music signal analysis, which is divided into three parts: traditional audio features, music related features and audio features for deep learning. It focuses on combing the meaning of various mainstream audio features and the application status of computing methods and music signal features in traditional machine learning methods and deep learning methods and summarizes the mainstream open source tools. Finally, the paper prospects the future development direction of the field of music features.