A Manual Bar-by-Bar Tempo Measurement Protocol for Polyphonic Chamber Music Recordings: Design, Validation, and Application to Beethoven's Piano and Cello Sonatas

Researcher Ignasi Sole has published a paper detailing a manual protocol for measuring tempo in historical polyphonic chamber music recordings that systematically outperforms automated computational tools. The method was specifically developed to address the failure of standard beat-detection software when applied to complex duo recordings like Beethoven's five piano and cello sonatas. Developed in collaboration with a VLSI design engineer, the protocol employs a cumulative timestamp architecture that prevents error accumulation across measurements while capturing subtle expressive timing phenomena—including rubato, fermatas, accelerandi, and ritardandi—that automated systems typically suppress or misread.

The protocol was rigorously applied to over one hundred movement-level recordings spanning from 1930 to 2012, generating a rich dataset of bar-level beats-per-minute (BPM) data with millisecond resolution. This data was then visualized using tempographs, histograms with spline-smoothed probability density functions, ridgeline plots, and combination charts. The paper argues that this manual approach is not a methodological step backward but a principled, necessary response to the intrinsic limitations of current AI and computational audio tools when faced with the specific acoustic and polyphonic challenges of historical recordings. The complete dataset and all analysis code have been made publicly available, providing a valuable resource for musicologists and audio researchers.