A living study of music generation, creator agency, and AI-native video culture

What happens when music creation becomes a conversation?

This project documents how artists and creative teams work with AI music generation tools, from early sketches and synthetic vocals to AI-assisted music videos and release-ready audiovisual experiments.

Read the project brief Participate as a creator

Research Questions

The project follows the friction points where creativity gets weird.

We are studying AI music generation as a practical creative environment: part instrument, part collaborator, part mirror, part legal and cultural headache machine.

Research Track 01

AI Music Generation Tools

We examine how creators use text-to-music, stem-generation, voice-synthesis, arrangement, and audio-to-audio systems in real creative workflows. The focus is not whether AI can make songs, but how these systems alter authorship, iteration, taste, labor, and decision-making.

Research Track 02

Creator Practice + Creative Agency

We document how musicians, filmmakers, editors, designers, and independent creators negotiate control when collaborating with generative systems. Our work studies prompt craft, revision behavior, attribution habits, aesthetic boundaries, and the moments where creators accept, reject, or reshape machine output.

Research Track 03

AI Music Videos + Synthetic Performance

We explore emerging relationships between generated music, AI-assisted video, synthetic performers, lyric visualization, motion systems, and platform-native release formats. The project treats AI music videos as cultural artifacts, not just technical demos.

Methodology

Built for messy, real-world creative workflows.

Instead of treating AI outputs as isolated artifacts, this project studies the full creative chain: intention, prompting, generation, rejection, revision, curation, editing, distribution, and audience framing.

Creator interviews and reflective practice journals

Workflow observation across music, video, and release pipelines

Comparative analysis of human-only, AI-assisted, and hybrid creative processes

Prompt, revision, and decision-path documentation

Ethical review of consent, attribution, dataset awareness, and creator disclosure

Ethics + Creator Rights

The study is designed around consent, clarity, and creative dignity.

AI music research can get sloppy fast. This project avoids treating creators as novelty witnesses for a technology story. Participants remain human authors with context, boundaries, histories, and rights.

Operating Principles

No fake certainty. Findings are labeled as provisional until reviewed.

No inflated numbers. Participation, outputs, and results are only published when verified.

Creators are treated as collaborators, not data points.

AI systems are studied as instruments, interfaces, and cultural forces.

Research outputs under development

The project will publish field notes, creator case studies, workflow diagrams, annotated tool experiments, and longer-form essays as material matures. Early posts will be clearly marked as working notes rather than final conclusions.

Research memoWorking NotesAI music is no longer a fringe demo category. Deezer reported in April 2026 that it receives nearly 75,000 fully AI-generated tracks each day, roughly 44% of daily uploads, while AI-generated listening remains only 1-3% of streams on its service. That gap makes the research question less about whether people can generate songs and more about how platforms, listeners, and creators sort abundance, trust, fraud, and attribution.Deezer Newsroom, April 2026
Justin Tyler Moore / TylerJayCreator Case StudiesThe initial case study follows Justin Tyler Moore as a creator working across memoir, platform design, AI-assisted songwriting, and tool education. Apple Books lists Burnt Echoes as a 2025 Vagabond Press memoir about rebuilding through creativity, resilience, and AI innovation. The handbook source material adds the practical side: a creator system built around brief, generate, repair, finish, with explicit attention to identity, stems, remastering, release archives, and rights.Apple Books + The Suno Creators Ultimate Handbook
Suno / Udio / workflow surfacesTool AnalysisThe handbook treats AI music tools as production environments, not magic buttons: start with the song job, choose the model for exploration, structure, or identity, then repair the smallest failing layer. Public product docs support the same shift toward creator-controlled workflows. Suno documents audio-upload creation and longer subscriber uploads, while Udio frames uploaded audio as material for extending, inpainting, sessions, remixing, and style transfer, with an explicit rights confirmation on upload.Handbook, Suno Help, Udio Help
Distribution, labels, and disclosureRelease CultureAI-native release culture now sits between normal music distribution and synthetic-media governance. IFPI reported recorded music trade revenues of US$29.6B in 2024, with streaming over US$20B, while also warning that unauthorized AI training threatens human artistry. YouTube separately requires disclosure for realistic or meaningful altered/synthetic content, including synthetically generated music. Release practice therefore has to cover metadata, rights, disclosure, and listener trust, not only audio quality.IFPI 2025 + YouTube Help
Deezer AI upload dataIFPI Global Music Report 2025 releaseRIAA Suno/Udio litigation statementU.S. Copyright Office AI copyrightability report releaseYouTube altered/synthetic content disclosureSuno audio uploadsUdio audio upload workflowBurnt Echoes on Apple Books
View fieldwork notes

Collaborate with the study

We are interested in speaking with musicians, producers, video artists, creative technologists, educators, labels, independent creators, and tool builders working with AI-assisted music or audiovisual systems.

Contact the research team