A step towards improving online music education by developing an AI tool that can recognise musical notation is described in the International Journal of Wireless and Mobile Computing. The work of Ting Zhang of the Academy of Arts at Shangluo University, Shaanxi, China, addresses a longstanding problem in digital music instruction, where the ability to recognise and interpret musical notation often falls short due to platform limitations. The research shows how image processing and machine learning can help online learners, allowing them to gain a richer, more accurate grasp of musical concepts.
Zhang has developed the Pulse-Coupled Neural Network (PCNN), an artificial neural network inspired by the workings of biological neurons, which “fire” in response to certain stimuli. Traditionally, online music education has relied on simplified digital representations of musical notation, leaving students without crucial guidance when attempting to understand the intricacies of symbols and musical structures.
The PCNN model focuses on improving the digital segmentation of musical symbols within an image of a musical score, for instance. By incorporating oblique spectral correction in the system, Zhang is able to break down the image into segments for precise differentiation between symbols. This allows even distorted representations of the music score to be analysed accurately, taking into account tilted symbols or misalignments.
The use of an optimized Convolutional Neural Network (CNN) for the image-recognition tasks makes the system efficient and accurate, giving it an up to 97 percent success rate.
For students, the enhanced notation recognition system could give them feedback in real-time even when no tutor is available for discussion. This system emulates face-to-face instruction, where instant feedback is usually available. The researchers saw notable improvements in student understanding of pitch and rhythm and in their grasping foundational music theory concepts.
Zhang, T. (2024) ‘Application of integrated image processing technology based on PCNN in online music symbol recognition training’, Int. J. Wireless and Mobile Computing, Vol. 27, No. 4, pp.369–380.
