摘要
提出一种新的基于音乐旋律的特征提取算法。该算法依据新的基音检测技术进行基音周期提取,进而得到每个音符的音高,然后使用一种新的线性伸缩方法进行旋律的匹配。实验结果表明这种新的音乐检索技术在哼唱检索中具备较高的准确性和检索效率。
Proposes a new feature extraction algorithm based on music melody.The algorithm is on the basis of a new fundamental tone detection tech?nology for pitch tracking,after obtaining the pitch of each note,a new linear scaling method is used to match this piece of melody.The ex?perimental results show that the new music retrieval technology enjoys higher accuracy and retrieval efficiency in humming retrieval.
作者
伍寅峰
张明
WU Yin-feng;ZHANG Ming(College of Information Engineering,Shanghai Maritime University,Shanghai 201306)
出处
《现代计算机》
2018年第6期3-7,共5页
Modern Computer
基金
国家自然科学基金(No.61603243)
关键词
哼唱检索
特征提取
音高序列
旋律匹配
Query by Humming
Feature Extraction
Pitch Sequence
Melody Matching