Research on the Design of Music Learning Singing System Based on Artificial Intelligence Feature Comparison
Yu Yu · 2024
This study aims to design and implement an AI-based music learning system focused on feature comparison. The system establishes rigorous criteria for vocal performance evaluation, utilizes Flash Media Server (FMS) for real-time web recording and storage, and employs sophisticated algorithms to extract pitch feature sequences from audio and MIDI files. Techniques such as sampling, pre-filtering, windowing, and fundamental frequency estimation are employed to generate pitch sequences, which undergo post-processing for refinement. Simultaneously, a melody extraction algorithm extracts the primary melody and its pitch features from MIDI files. The design emphasizes precise extraction and comparison of pitch features to ensure users receive efficient and accurate feedback. Finally, the system's efficacy is validated through practical cases, analyzing its performance in real-world applications and showcasing its vast potential in music education.