Research Project

Deckless

Exploring what happens when AI truly understands music. A Large Music Matching Model for contextual audio intelligence.

What if AI could hear music the way a DJ does?

Not just tempo and key, but vibe. Energy. The moment where two tracks click together and become something new. We're building the foundational model to find out.

The proof of concept: an AI mashup creator that discovers compatible songs, beat-matches automatically, and helps anyone build DJ-quality sets—without the decks.

Building contextual music understanding

🎵

Deep Audio Analysis

Understanding structure, energy curves, and sonic texture beyond simple BPM and key detection.

🔗

Compatibility Matching

Predicting which songs will work together—and why. Building the intuition DJs develop over years.

Intelligent Blending

Auto beat-matching, key adjustment, and transition suggestions that actually sound good.

🎛️

Creative Tools

Tweaking, polishing, and set-building tools that turn rough mashups into finished mixes.

From links to full DJ sets

1

It starts with a song. Or two. Or three.

Drop a YouTube, SoundCloud, or Spotify link. We extract and analyze the audio.

2

Discover matches

AI suggests songs that would mash well—based on structure, energy, and sonic compatibility.

3

Auto-blend

Beat-matching, key alignment, and transition points are handled automatically.

4

Tweak & polish

Fine-tune the mix with editing tools. Adjust levels, transitions, and effects.

5

Build the set

Chain mashups together into a full DJ set. Export and share.

Interested in the research?

We're looking for collaborators, early testers, and people who think deeply about music and AI.

Start a conversation