Article: Transformer-based real-time automatic error annotation for piano performance Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.53 pp.24 - 46 Abstract: This research tackles the pressing challenge of real-time automatic error detection in piano performance, a task where conventional approaches often propagate inaccuracies due to the decoupling of audio-score alignment and error identification. This paper introduce the DiffAlignTransformer framework, which incorporates a differentiable dynamic programming mechanism to jointly learn probabilistic notelevel alignment and error classification within a hierarchical crossmodal encoder. Evaluated on the Vienna Synchronous Library dataset using a leaveoneperformerout validation strategy, the model attains an overall F1score of 0.872, exceeding the strongest baseline by 6.0%, with marked gains in onset (7.2%) and offset (8.1%) error recognition. Inference requires only 78 milliseconds per second of audio, satisfying strict realtime constraints. These outcomes confirm that our method successfully resolves the intertwined alignment-detection problem and delivers precise, instantaneous feedback for piano pedagogy. Inderscience Publishers - linking academia, business and industry through research

Title: Transformer-based real-time automatic error annotation for piano performance

Authors: Yue Gu

Addresses: School of Music, Jianghan University, Wuhan, 430000, China

Abstract: This research tackles the pressing challenge of real-time automatic error detection in piano performance, a task where conventional approaches often propagate inaccuracies due to the decoupling of audio-score alignment and error identification. This paper introduce the DiffAlignTransformer framework, which incorporates a differentiable dynamic programming mechanism to jointly learn probabilistic notelevel alignment and error classification within a hierarchical crossmodal encoder. Evaluated on the Vienna Synchronous Library dataset using a leaveoneperformerout validation strategy, the model attains an overall F1score of 0.872, exceeding the strongest baseline by 6.0%, with marked gains in onset (7.2%) and offset (8.1%) error recognition. Inference requires only 78 milliseconds per second of audio, satisfying strict realtime constraints. These outcomes confirm that our method successfully resolves the intertwined alignment-detection problem and delivers precise, instantaneous feedback for piano pedagogy.

Keywords: piano performance assessment; error detection; differentiable alignment; cross‑modal transformer; real‑time feedback.

DOI: 10.1504/IJICT.2026.153713

International Journal of Information and Communication Technology, 2026 Vol.27 No.53, pp.24 - 46

Received: 23 Jan 2026
Accepted: 27 Feb 2026
Published online: 21 May 2026 *

Title: Transformer-based real-time automatic error annotation for piano performance

Keep up-to-date