摘要： 在音乐流派分类过程中，音乐流派局部特征与整体特征不一致时，通常采用的局部特征投票取最大的方法(MaxVote)在音频片段流派分类精度不高，而流派特征分布比较均衡时分类结果不合理。针对以上问题，该文提出基于音乐片段流派分布特征的神经网络投票机制(NNVote)和结合高层音乐节奏特征的RhythmNNVote投票方法。实验结果表明，NNVote方法在7个流派上的分类总精度达到68.9%，较MaxVote提高将近10%。

关键词: 音乐流派分类, 音乐信息检索, 音频特征提取

Abstract: During the process of music genre classification, when dealing with inconsistency between the local genre features and the global genre attribute, the MaxVote method which determines the whole genre by the dominate local genre is not so reasonable when the local feature is distributed uniformly. The paper proposes a method called NNVote based on clip genre distribution feature and a method called RhythmNNVote which combines the high level rhythm feature and low level features. In the experiment, the NNVote method total classification accuracy reaches 68.9% on 7 genres which improves about 10% compared with MaxVote.

Key words: music genre classification, music information retrieval, audio feature extraction

中图分类号: 

• TP18