National Gallery Mixtape

National Gallery Mixtape treats paintings as ingredients for a small generative sound studio. Arrange up to six works, add stickers that nudge mood or musical style, then either play the sequence as a score or blend its parts with mixer faders. It is closer to a curious synesthetic toy than an audio guide.

What is National Gallery Mixtape?

The premise is wonderfully literal: if a painting had to become sound, what would it hand to the speakers? Choose works from the National Gallery's digitized collection and place them in order, then use stickers to push the result toward a feeling, instrument, or genre. Behind the scenes, the project says Gemini turns each image into a description and MusicFX DJ uses that text to generate audio. Scores mode makes the pictures behave like stops along a musical route; Mixer mode lets them overlap, so a serene landscape can rub shoulders with a severe portrait instead of waiting its turn.

What you can do there

• Arrange up to six National Gallery paintings on a score
• Steer the soundtrack with mood, style, or instrument stickers
• Layer generated sounds with virtual mixer faders

Why we picked it

This experiment gives generative music a set of objects worth arguing with. The fun is not discovering a correct soundtrack for a masterpiece—there is no such thing—but comparing how the mix changes when you swap one canvas, reorder the score, or lean on a mood sticker. It makes looking and listening feel like the same bit of play.

How to get the most from it

Start with two paintings that clash in color or atmosphere, listen once, then add a single sticker so you can hear what changed. Build toward six only after the basic relationship makes sense, and move to Mixer mode when you want to balance layers rather than follow a sequence.

Good to know

Audio is the point, and the project hub recommends a larger screen. The public pages do not spell out model versions, generation limits, saving or download behavior, music reuse rights, privacy details, accessibility support, or how faithfully the system's descriptions reflect each artwork.

Who made it, and when?

The National Gallery, London and Google Arts & Culture Lab is the credited creator or organization. The earliest supported launch date we found is November 13, 2024.