An interactive research presentation
This project asks three questions about music that are difficult to answer by listening to one piece alone. It follows poetic line endings through Monteverdi’s madrigals, recurring phrases through hundreds of jazz solos, and changes in hit songs across seventy-two years. Each study begins with something we can hear, finds a way to count it, and checks whether the pattern holds beyond the example that first caught our attention.
Open on your phoneScan to visit
music.tylersotomayor.com
We start with one madrigal, one jazz phrase, and one hit song—then ask what changes when we compare hundreds more.
Each movement has its own background track. Playing an example lowers the soundtrack; it returns when the clip ends.Why compute music?
Many musicological arguments begin with something a listener notices: a charged word, a familiar phrase, the lift of a chorus. Digital scores and transcriptions let us check whether that moment is exceptional or part of a larger habit.
Counting gives us firmer ground for interpretation. We can ask how often a pattern occurs, where it appears, and whether it still appears when we change the rules used to find it.
We compare how Monteverdi sets expressive words and ends lines of poetry.
We track short interval patterns across transcribed solos and compare each player’s vocabulary.
We revisit a published trend over a longer period, then compare verses and choruses within the same songs.
Our approach Start with something audible, decide how to measure it, check the rest of the corpus, and see how the result changes under different choices.
In Monteverdi, the notes were fixed in a score.
↓The jazz study compares musicians.
↓Where could this research go next?
The measurements can be reused in other repertories, but they need more historical evidence and more listening before they can support broader claims.
We could repeat the boundary analysis in sacred music, opera, and later song, then compare the written score with recorded performances.
Needed next: more composers, translations, and score-aligned recordings.Dates, teachers, bands, and cities would help separate a widespread convention from a phrase that may have passed between musicians.
Needed next: social histories, session personnel, and phrase-aligned audio.Song features could be linked to radio formats, label concentration, streaming incentives, and changes in release strategy. That would let us study specific shifts rather than treating the calendar year as a cause.
Needed next: platform and industry data, plus a causal research design.A listening study could test whether people hear the poetic boundary, recognize a transposed interval pattern, or identify a chorus from melody alone.
Needed next: listener responses and carefully matched audio examples.Digital archives let us revisit familiar stories at a larger scale. They also inherit the limits of what was preserved and encoded, so the makeup of the corpus must remain part of the argument.
What did the three methods show?
Some meanings receive recurring treatment, and poetic line endings consistently receive more time.
Players share recurring interval patterns, but the strength of a connection depends on how a lick is defined.
A trend that looks clear over thirty years can bend or disappear when we add four more decades.
We began by listening for a detail. The data told us where else to listen—and, just as often, when our first impression needed revision.Return to the beginning ↑