Musical sound recognition

• Main Authors: Foo, Say Wei, Ng, Chen Hwi
• Other Authors: School of Electrical and Electronic Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/90747; http://hdl.handle.net/10220/4705
• Institution: Nanyang Technological University

In this paper, two different methods to implement one particular aspect of music transcription – musical note recognition – are presented. Both methods make use of existing algorithms in the form of the Constant-Q Transform (CQT) and the Discrete Wavelet Transform (DWT). Each existing algorithm is modified or extended so that a musical note recognition algorithm can be implemented on computer. Two main principles behind each method are peak detection using suitable thresholds and check for presence of harmonics in the process of note identification. The CQTbased method has a higher degree of accuracy, as it is able to resolve up to four simultaneous notes played on a guitar. The DWT-based method is less accurate but some suggestions are given as to how the algorithm can be further modified for improved performance.