Back to portfolio

Algorithmic Ear

How Machines Listen

Algorithms

Experimental tool exploring algorithmic perception

01. Figures

Interface overview — The Listening Module. Establishes environment.

Fig. 1

Interface overview — The Listening Module. Establishes environment.

Picker interface — Selection tool. Shows decision mechanism.

Fig. 2

Picker interface — Selection tool. Shows decision mechanism.

Algorithmic trace — Greedy optimization path visualization. Shows how logic operates.

Fig. 3

Algorithmic trace — Greedy optimization path visualization. Shows how logic operates.

Slider interaction — Emotional weighting system. Shows user control.

Fig. 4

Slider interaction — Emotional weighting system. Shows user control.

Features — Key capabilities of the algorithmic perception system. Supplemental explanation.

Fig. 5

Features — Key capabilities of the algorithmic perception system. Supplemental explanation.

How it works — Explanation of the tool's functionality. Supplemental explanation.

Fig. 6

How it works — Explanation of the tool's functionality. Supplemental explanation.

Next song — Algorithm recommendation result. Outcome view.

Fig. 7

Next song — Algorithm recommendation result. Outcome view.

The future — Forward-looking vision for algorithmic perception. Closing visual.

Fig. 8

The future — Forward-looking vision for algorithmic perception. Closing visual.

02. Abstract

A prototype that visualizes recommendation algorithms. Spotify converts audio into numerical features—danceability, energy, valence—and this tool uses a greedy algorithm to show how feature engineering shapes what we hear and how we discover music.

03. Research Question

What happens when recommendation algorithms are visible? How do users interact with algorithmic logic when they can see features driving song selection? What does this reveal about computational efficiency versus musical discovery?

04. Hypothesis

When users manipulate which features drive recommendations, they discover that simple algorithms create loops—locally optimal but globally suboptimal. The greedy algorithm shows how feature-based optimization prioritizes immediate similarity over diverse discovery.
ElementPurpose / Specification
Color Palettegray, black (#6b7280, #000000)
TypographyInter, Monospace, Sans-serif, Serif
Design PrinciplesMinimalism

Fig. 0 Visual system schematic with design rationale.