Evaluation of National Music Performance by Integrating Attention Mechanism and Music Theory Rules

Abstract

The evaluation of ethnic music performances is crucial for music education and cultural heritage preservation. This study proposes an intelligent evaluation model that integrates music theory principles and attention mechanisms (AMs). This model aims to enhance the objectivity and accuracy of assessments. The model uses a complex number convolutional neural network (CN-CNN) to process musical audio signals and extract spectral features. It also incorporates an AM-enhanced long short-term memory (LSTM) algorithm to enhance its ability to extract features. This effectively addresses dynamic pitch and rhythmic variations in improvisation. The results demonstrated that compared to traditional methods, the model exhibited superior training efficiency and convergence performance, achieving 98.01% accuracy and an F1 score of 0.92. In practical applications, the model demonstrated high accuracy in melody recognition and harmonic evaluation, showing remarkable consistency with professional auditory assessments. This research offers a new way to objectively and precisely evaluate ethnic music performances. This method contributes to music education and cultural preservation.