音乐信息检索下的乐器识别综述

Review of Musical Instrument Recognition in Music Information Retrieval

高效精准的乐器识别技术可以有效地推动声源分离、音乐识谱、音乐流派分类等研究的深入发展,可广泛应用于播放列表生成、声学环境分类、乐器智能教学和交互式多媒体等众多领域。近年来,随着乐器识别研究的不断推进,乐器识别系统在性能上有了大幅提高,但依旧存在着部分乐器难以识别、乐器音频特征提取较为困难、复音乐器识别精准度较低等诸多问题,如何借助人工智能技术对乐器进行高效精准的识别成为当前研究的热点和难点。针对当前研究现状,从乐器识别常用音频特征、乐器识别模型及方法和常用数据集三个方面进行综述,并对当前研究中存在的局限性和未来发展趋势进行总结,为乐器识别研究提供一定的借鉴参考。

Efficient and accurate instrument recognition technology can effectively promote the in-depth development of sound source separation,music spectrum recognition,music genre classification and other research,and can be widely used in many fields,such as playlist generation,acoustic environment classification,instrument intelligent teaching and interactive multimedia.In recent years,with the continuous advancement of musical instrument recognition research,the performance of musical instrument recognition system has been greatly improved,but there are still many problems,such as difficult recognition of some musical instruments,difficult extraction of musical instrument audio features,and low recognition accuracy of complex musical instrument.How to identify musical instruments efficiently and accurately with the help of artificial intelligence technology has become the focus and difficulty of current research.According to the current research status,this paper summarizes the commonly used audio features,musical instrument recognition models and methods and commonly used data sets of musical instrument recognition,and summarizes the limitations and future development trend of the current research,so as to provide some reference for the research of musical instrument recognition.