Development and application of piano accompanying system based on New Fingerprint algorithm

Abstract

In recent years, due to the rapid development of online teaching, intelligent piano accompaniment has emerged. However, the relevant applications on the market are relatively low in intelligence. It is difficult to identify the pieces played by the user, which increases operational complexity. To solve the audio retrieval when users are learning online piano, a New Fingerprint (NF) is designed based on the button behavior of users while playing the piano. Taking NF algorithm as the core, combined with music signal analysis technology, automatic piano transcription technology and alignment technology, an intelligent piano audio analysis module is constructed. Finally, a complete piano accompaniment system is established based on spectrum reading mode, practice mode, and piano performance scoring mode. The research results showed that the average precision of NF algorithm was 98.76 %, the average recall was 87.84 %, and the average F1 value was 98.53 %. The average accuracy value of the piano accompaniment system based on NF algorithm was 97.53 %, and the accuracy N value was 1.2. In practical application, 88 % of users were very willing to learn from the piano accompaniment system, and 45 % of users were very satisfied, and 28 % of users were relatively satisfied. To sum up, the proposed NF algorithm has excellent performance. The piano accompaniment system based on NF algorithm is suitable for actual piano learning.