The Sounds of Manhattan


Creating music out of the architecture of Manhattan.

The question

Can we create music out of architecture?

I've been wondering this ever since I became interested in architecture during high school. I'd walk past sky scrapers in New York and wonder if you could fly a helicopter over it, then turn the entire city into a series of notes.

Inspired by the theramin, I wondered if you could shoot a beam down from a helicopter and get a frequency back based on distance.

My beautiful helicopter.

I eventually realized that using the tallest building as a reference point was probably not the best way to get a diverse range of sounds, so I switched to taking the ratio of each building's height and the median building height.

The results

I created this entire experiment  in Python and used a public Shapefile of Manhattan to parse through the data.

The Jupyter Notebook is available here.

Once I mathematically defined the relationships between frequency and note duration, I was able to produce the MIDI files associated with the processed data.

Testing on Wall Street

In the meantime while I don't have a proper data visualization (New Years is busy...) here is a recording of the MIDI roll in logic. I exported the MIDI file, threw on some synthesizers, and dropped everything 2 octaves.


Wave Plot

Musical Score



Notice how the frequency ranges look eerily like buildings.

What's next?

I figure that if you can go from architecture -> music, you can also go from music -> architecture. Soon I'll expand this project by creating 3D generated models of cities based off of different music-- ranging from Beethoven and Mozart to Daft Punk. Perhaps cities in the future will be "culturally intelligent" and be able to rearrange themselves based on large events and themes.