Construction of Generative AI Piano Improvisation Accompaniment Teaching System Based on Fine-Tuned Music Large Model
Main Article Content
Abstract
Current approaches to piano improvisation accompaniment teaching are hampered by such aspects as delayed feedback or inadequate personalized instruction and digital tools are also limited technically in terms of music theory constraints and realtime interaction. This research builds a generative artificial intelligence teaching system using a large-scale music model fine-tuned. The system front-end obtains the student playing data through the Web MIDI API and extracts the features of non-stationary tempo and dynamic using the temporal difference normalization algorithm and exponential moving average algorithm respectively. The main accompaniment engine is based on the large-scale model ChatMusician and is fine-tuned with datasets using LoRA to generate the accompaniment in a grammatically correct way by applying harmonic constraints by implementing the Top-p sampling strategy. The back-end is dedicated to the integration of the dynamic time warping algorithms and visualization charts to give multi-dimensional quantitative feedback. The results of the experiments after 8 weeks of teaching indicated that the error rate for the experimental group was reduced to 7.6%, fluency score was 86.7 and the system usability scale score was 82.4. It has the ability to effectively reduce the harmonic grammar error in this system, enhance the playing fluency and learning independence of music learners, and offer a practical technical solution for intelligent music education.
Downloads
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
References
J. Yuan, “Construction of AI curriculum system for piano and improvisation accompaniment based on knowledge graph,” Heilongjiang Science, vol. 16, no. 13, pp. 87–89, 2025, doi: 10.3969/j.issn.1674-8646.2025.13.026.
J. Wang, “A Study on the Three Elements of Piano Improvisation Accompaniment Teaching,” Journal of Xinjiang Arts Institute, vol. 21, no. 4, pp. 33–39, 2023, doi: 10.3969/j.issn.1672-4577.2023.04.006.
J. Liu, “Exploration of piano teaching models and development paths in higher vocational music education,” Journal of Yanbian Teachers College, vol. 40, no. 2, pp. 41–45, 2026, doi: 10.3969/j.issn.1673-4564.2026.02.008.
X. Ke, “Innovation and practical exploration of piano teaching model based on constructivist theory,” Journal of Jiamusi Vocational College, vol. 42, no. 5, pp. 187–189, 2026, doi: 10.3969/j.issn.2095-9052.2026.05.062.
M. Guo, “Construction of a piano improvisation accompaniment teaching platform based on Logic Pro software,” Microcomputer Applications, vol. 38, no. 7, pp. 154–156 + 164, 2022, doi: 10.3969/j.issn.1007-757X.2022.07.043.
K. Isarathikul, P. Promsukkul, and N. O. Tayrattanachai, “THE GUIDELINES FOR INTERMEDIATE CLASSICAL PIANO INSTRUCTION OF THE PIANO ACADEMY OF BANGKOK,” Mahidol Music Journal, vol. 7, no. 2, pp. 1–19, 2024. THE GUIDELINES FOR INTERMEDIATE CLASSICAL PIANO INSTRUCTION OF THE PIANO ACADEMY OF BANGKOK - Mahidol University.
P. Madapatha, “Examining the Impact of Individual and Group Piano Instruction on Students’ Learning: Teachers’ Perceptions,” International Journal of Teaching and Learning in Higher Education, vol. 36, no. 1, p. 14, 2025, doi: 10.7771/1812-9129.1043.
C. H. Choi and C. K. Lim, “Digital Innovations in Group Piano Instruction,” Journal of the Association for Technology in Music Instruction, vol. 5, no. 2, p. 20, 2025. “Digital Innovations in Group Piano Instruction” by Chee Hyeon Choi and Chan Kiat Lim.
X. Chen and T. Phongsatha, “The Application of Mobile Blended Learning in Piano Instruction for Non-Piano Major Students at Guangdong University of Petrochemical Technology,” International Journal of Sociologies and Anthropologies Science Reviews, vol. 4, no. 4, pp. 471–486, 2024, doi: 10.60027/ijsasr.2024.4477.
X. Liu and T. Phongsatha, “The Efficacy of Blended Learning via Video Conferencing for Piano Instruction: A Case Study at Yunnan University,” International Journal of Sociologies and Anthropologies Science Reviews, vol. 5, no. 4, pp. 875–888, 2025, doi: 10.60027/ijsasr.2025.6931.
I. Arshinova, “Elements of piano education and the evolution of piano pedagogy,” Journal for the Interdisciplinary Art and Education, vol. 3, no. 2, pp. 81–97, 2022. Elements of piano education and the evolution of piano pedagogy - Journal for the Interdisciplinary Art and Education.
A. Akgul, “An analysis on piano education in institutions training music teachers in the context of application, teaching experience and pre-school period,” Kıbrıslı E˘gitim Bilimleri Dergisi, vol. 17, no. 2, pp. 323–342, 2022. CEEOL - Article Detail.