Figure 1
Distribution of genres (top) and years (bottom) in the duple‑meter‑only sub‑dataset used for the microtiming analysis.
Figure 2
Distribution of the ratios of the vocal stem root mean square (RMS) energy to the full mix RMS energy in the data. A threshold of 0.08 was used to discard nonvocal clips.
Figure 3
Visualizing one 30‑s song excerpt using the carat microtiming tool. Top: Each dot represents a vocal onset distributed among four equal subdivisions of a bar (1, 2, 3, 4), Bottom: Onsets distributed among four equal subdivisions of a beat (N, E, &, A). The horizontal axis shows the timing deviation (in beat‑normalized time) relative to the mean onset location for each metrical position. In the vertical axis, each row corresponds to an individual onset event from the excerpt, allowing all onsets to be displayed simultaneously. At the bottom, a histogram of the onset locations is depicted. We fit a normal distribution to the onsets in each subdivision; the mean is shown as a vertical dotted line and as a percent, and the standard deviation is shown as horizontal ‘arms.’ The straight‑timed (left) plot is an illustration of MSD track ID: TRUUJZA128F931525C, Mr. Brown by Styles of Beyond.6 The deviated example (right) illustrates MSD track ID: TRIOFIX128F93156AD, Sunndal Song by The Apples in Stereo.7
Figure 4
Examples of vocal and mix spectrograms with overlaid estimated onset, beat, and downbeat annotations, comparing a straight‑timed performance (left) with one showing microtiming deviations (right). The straight‑timed (left) plot is an illustration of MSD track ID: TRUUJZA128F931525C, Mr. Brown by Styles of Beyond.6 The deviated example (right) illustrates MSD track ID: TRIOFIX128F93156AD, Sunndal Song by The Apples in Stereo.7
Figure 5
Microtiming in beats (i.e., quarter notes) in each of the 10 genres across the dataset. Means are shown with boxes representing the interquartile range, the horizontal line inside the box marks the median, and the whiskers extend to 1.5 times the interquartile range.
Figure 6
Microtiming of each beat as a function of year globally. Each dot represents a song. The red line represents the predicted slope with 95% confidence intervals. The green diamond and ribbon represent the mean per year and the standard error.
Figure 7
Microtiming in beat subdivisions (i.e., 16th notes) across 10 musical genres in the dataset, for all four subdivisions. Each box represents the interquartile range (25th–75th percentile), the horizontal line inside the box marks the median, and the whiskers extend to 1.5 times the interquartile range.
Figure 8
Microtiming of each beat subdivision as a function of year globally. Each dot represents a song. The red line represents the predicted slope with 95% confidence intervals. The green diamond and ribbon represent the mean per year and the standard error.
Figure 9
Microtiming of each beat subdivision pair as a function of year globally. Each dot represents a song. The red line represents the predicted slope with 95% confidence intervals. The green diamond and ribbon represent the mean per year and the standard error.
Table 1
Table depicting the number of songs in each genre and the average of the per‑song vocal onset–to‑beat ratios, with standard deviations.
| Genre | Songs | Onset Density | |
|---|---|---|---|
| Avg. | Std. | ||
| Rap | 5,373 | 2.06 | 1.11 |
| Reggae | 3,058 | 1.09 | 0.69 |
| RnB | 5,533 | 1.04 | 0.44 |
| Latin | 2,136 | 0.95 | 0.43 |
| Pop | 12,475 | 0.83 | 0.36 |
| Country | 3,096 | 0.77 | 0.31 |
| Elec. | 5,878 | 0.70 | 0.60 |
| Rock | 40,079 | 0.67 | 0.37 |
| Metal | 3,547 | 0.61 | 0.30 |
| Punk | 2,555 | 0.59 | 0.24 |
Table 2
Table depicting the number of songs in each demi‑decade and the average of the per‑song vocal onset–to‑beat ratios, with standard deviations.
| Demi‑Decade | Songs | Onset Density | |
|---|---|---|---|
| Avg. | Std. | ||
| 1965–1969 | 1,072 | 0.69 | 0.25 |
| 1970–1974 | 1,701 | 0.78 | 0.31 |
| 1975–1979 | 2,491 | 0.72 | 0.28 |
| 1980–1984 | 3,721 | 0.69 | 0.32 |
| 1985–1989 | 5,103 | 0.72 | 0.39 |
| 1990–1994 | 7,922 | 0.81 | 0.50 |
| 1995–1999 | 12,303 | 0.86 | 0.55 |
| 2000–2004 | 21,158 | 0.87 | 0.55 |
| 2005–2010 | 32,887 | 0.84 | 0.71 |