| Author(s) |
|
|---|---|
| Abstract |
Using the Lempel-Ziv algorithm (1977), Morris (2017) demonstrated that pop song lyrics have become increasingly repetitive over time, measured by text compressibility ratio, and that top 10 songs were more repetitive than the rest of the dataset. Humdrum encodes musical data as text, permitting compressing the information using the same method. This study tests two hypotheses: 1) pop song melodies have similarly become increasingly repetitive over time as a measure of compressibility ratio, and 2) top 10 melodies are more repetitive than less popular songs. This study tests its hypothesis using a new dataset: 718 MIDI encodings of the top 20 Billboard year-end popular songs from 1956-2015 scraped from the web. Using Humdrum, melodies are extracted from complex textures, turned into text strings, and subjected to the Lempel-Ziv algorithm, resulting in a percentage of the file compressed for each song. Consistent with H1, pop song melodies have become increasingly compressible over time (R=+.23, suggesting that popular song melodies are more repetitive now than in the past. However, amount of melodic compressibility is not statistically significant (p=.070) between songs in the Top 10 vs. from chart position 11-20. There is also not a significant interaction between chart position and year in predicting compressibility, although it is close to the significance level (p=.0502), suggesting that the effect of chart position on melodic repetitiveness may have less impact today than in earlier years. In fact, the range of compressibility is much narrower after about 2001.
|
| Format | |
| Contributor(s) |
Faculty Mentor
Joshua Albrecht |
| Modified Abstract |
Using the Lempel-Ziv algorithm, this study tests two hypotheses: 1) pop song melodies have similarly become increasingly repetitive over time as a measure of compressibility ratio, and 2) top 10 melodies are more repetitive than less popular songs. This study tests its hypothesis using a new dataset: 718 MIDI encodings of the top 20 Billboard year-end popular songs from 1956-2015 scraped from the web. Using Humdrum, melodies are extracted from complex textures, turned into text strings, and subjected to the Lempel-Ziv algorithm, resulting in a percentage of the file compressed for each song. Consistent with H1, pop song melodies have become increasingly compressible over time (R=+.23, suggesting that popular song melodies are more repetitive now than in the past. However, amount of melodic compressibility is not statistically significant (p=.070) between songs in the Top 10 vs. from chart position 11-20. |
| Permalink | https://oaks.kent.edu/ugresearch/2020/art/melodies-pop-measuring-melodic-repetitiveness-throughout-billboard-era |
Sanchez, J. Melodies that Pop: Measuring Melodic Repetitiveness Throughout the Billboard Era. https://oaks.kent.edu/node/10298
Sanchez, Joshua. n.d. “Melodies That Pop: Measuring Melodic Repetitiveness Throughout the Billboard Era”. https://oaks.kent.edu/node/10298.
Sanchez, J. Melodies That Pop: Measuring Melodic Repetitiveness Throughout the Billboard Era. https://oaks.kent.edu/node/10298.