Towards Machine Learning of Expressive Microtiming in Brazilian Drumming

We have used supervised machine learning to apply microtiming to music specified only in terms of quantized note times for a variety of percussion instruments. The output of the regression schemes we tried is simply the microtiming deviation to apply to each note. In particular, we trained Locally Weighted Linear Regression / KNearest-Neighbors (LWLR/KNN), Kernel Ridge Regression (KRR), and Gaussian Process Regression (GPR) on data from skilled human performance of a variety of Brazilian rhythms. Although our results are still far from the dream of inputting an arbitrary score and having the result sound as if expert human performers played it in the appropriate musical style, we believe we are on the right track. Evaluating our results with cross-validation, we found that the three methods are quite comparable, and in all cases the mean squared error is substantially less than the mean squared microtiming of the original data. Subjectively, our results are satisfactory; the applied microtiming captures some element of musical style and sounds much more expressive than the quantized input.

[1]  J. Sundberg,et al.  Musical performance. A synthesis‐by‐rule approach , 1981 .

[2]  Daniel P. W. Ellis,et al.  Eigenrhythms: Drum pattern basis sets for classification and generation , 2004, ISMIR.

[3]  Thomas G. Dietterich.  What is machine learning? , 2020, Archives of Disease in Childhood.

[4]  V. I. J. Ay.  Embodied Mind, Situated Cognition, and Expressive Microtiming in African-American Music , 2002 .

[5]  Geoffrey E. Hinton,et al.  Evaluation of Gaussian processes and other methods for non-linear regression , 1997 .

[6]  Henkjan Honing,et al.  From Time to Time: The Representation of Timing and Tempo , 2001, Computer Music Journal.

[7]  Carl Haakon Waadeland.  “It Don’t Mean a Thing If It Ain’t Got That Swing” – Simulating Expressive Timing by Modulated Movements , 2001 .

[8]  Gerhard Widmer,et al.  Large-scale Induction of Expressive Performance Rules: First Quantitative Results , 2000, ICMC.

[9]  Walter Gerard Busse.  Toward objective measurement and evaluation of jazz piano performance via MIDI-based groove quantize templates , 2002 .

[10]  Gerhard Widmer,et al.  Machine Discoveries: A Few Simple, Robust Local Expression Principles , 2002 .

[11]  Joseph T. Wells.  Timing Is of the Essence , 2001 .

[12]  Maarten Grachten,et al.  A case based approach to expressivity-aware tempo transformation , 2006, Machine Learning.

[13]  Eric Clarke,et al.  Rhythm and Timing in Music , 1999 .

[14]  Jeff A. Bilmes,et al.  Timing is of the essence : perceptual and computational techniques for representing, learning, and reproducing expressive timing in percussive rhythm , 1993 .

[15]  David A. Jaffe.  Ensemble Timing in Computer Music , 1984, ICMC.

[16]  Anders Friberg,et al.  Swing Ratios and Ensemble Timing in Jazz Performance: Evidence for a Common Rhythmic Pattern , 2002 .