Design and performance analysis of a blind source separation algorithm for piano music signals

As the modern music industry continues to advance, musical signal separation technology has become essential to piano music production, analysis, and education. However, during actual piano performances, the number of observed signals may be greater than, equal to, or less than the number of source signals. This involves both determined and underdetermined blind source separation scenarios. In view of this, the study proposes an intelligent music signal blind source separation algorithm model. The model consists of improved independent component analysis and sparse component analysis based on fuzzy C-mean algorithm. The study validated the proposed model. According to the experimental findings, enhanced independent component analysis produced an average accuracy of 93.45% when noise was present and 97.96% similarity between the separation signal and the source signal when noise was absent. Compared with the traditional model, it improved the accuracy of piano music signal separation. In addition, the average time-domain similarity and frequency-domain similarity of sparse component analysis were 98.72% and 99.27%, respectively. The similarity as well as the accuracy of dealing with different music styles ranged from 97.12% to 99.05%. In processing mixed audio with different numbers of source signals, the average similarity of sparse component analysis reached 98.86% and 98.38%, respectively, which outperformed other comparison models. In summary, the proposed intelligent music signal blind source separation model provides a strong support for piano music visualization and has certain research significance for piano music teaching and music creation.