(back)
Brushed
Brushed

Date

December 2024

Type

music visualizer

Made with

OpenAI API, Genius and LRC LIB API, p5.sound, p5.js, d3.js

A web platform that creates paintings based on songs, leveraging OpenAI for sentiment analysis based on the song's lyrics, and p5.sound for music analysis.

Concept

Brushed emerged from my wish to combine music with artistic process. I aimed to combine different aspects of music with natural brush textures to create a data-visualization project that generates a painting in real time, where art interacts with music.

Project image

Process(Code)

The user submits an audio file of their chosen song, alongside the song's title and artist name. The data is sent to the Genius and LRC LIB APIs to generate timestamped lyrics. This information is then submitted to the OpenAI API, which separates the song into 5-10 main sections, each with its own dataset.

Project image

process (data visualization)

Leveraging p5.sound and the Web Audio API, I mapped each painting into different fields dictated by different pitches found in the song, creating Voronoi fields with D3.js.

Project image

process (brushes)

I created a mini-library with custom brush textures, which were then used to create the paintings.

Project image
Project image
Project image

example output: "Hannah Hunt" by Vampire Weekend

PS.: this is my favorite song :) Unmute the video to follow along with the song.

Project image