Fig 1.
Upper panel: Audio signal of a short clip of the song presented as a spectrogram.
The bright branches at high frequencies correspond to the hi-hat beats. Lower panel: Cross section of the spectrogram with the envelope, amplitude threshold (dashed line), and the detected onset times (crosses).
Fig 2.
(a) Detected 16th note intervals in the song. (b) Detected onset amplitudes. (c-d) Corresponding statistical distributions of (a) and (b), respectively. The distribution of the intervals resemble a Gaussian distribution with a mean value 157 ms and a standard deviation of 8.7 ms.
Fig 3.
Drift of the sixteenth note pulse, i.e., the deviation from an imaginary metronome during the song.
A, B1, B2, and C mark different parts of the song (see text).
Fig 4.
Results from the DFA (left panel) and power spectral analysis (right panel) for the fluctuations of the interbeat intervals (a, b) and the beat amplitudes (c, d).
Fig 5.
Poincaré plot (return map) of the interbeat intervals that map the i:th interval to the following one.
The data shows negative correlations with Pearson coefficient −0.48.
Fig 6.
(a) 16th note interbeat intervals during the first 10 bars in “I Keep Forgettin”’. Each bar consists of 16 hits, and the corresponding intervals are marked at half-integer values. The interval between the 15th and 16th note is missing due to open hi-hats at the end of almost all bars (see text). The circles and squares correspond to odd (1, 3, …) and even (2, 4, …) bars, respectively. (b) Same as (a) but for the hit amplitudes. Again, the 16th hit is missing due to the open hi-hat—see the drum score in Fig 7.
Fig 7.
Drum score of bars 9-12 in “I Keep Forgettin”’.
The 16th notes are marked on the upmost line. The (clearest) accents are marked by the symbol “>”, and the open hi-hat is marked by a symbol “o” on top of the 8th notes. The score can be compared to the detected 16th note intervals and amplitudes in Fig 6. The drum score of bars 1-4 is identical apart from the replacement of the snare by a cross-stick rimshot.