Research

GROUNDED IN RESEARCH

Established 2024

From Labs
To Legends

University AI Music Research Serving Artists

‍

Our 2026 initiative brings researchers together to study Gen-AI music models, attribution & benchmarking, uncovering songs used in AI development, and tools that identify copyrighted material in AI-generated works.

2026 RESEARCH

CLAIMS

EVIDENCE

01.

Source:

We uncover which songs and recordings are used to train and advance foundational AI and machine-learning models.

02.

Model:

With NYU, UC Riverside and Cal Poly, we study generative music models built to carry provenance from the start and explain what shaped their outputs.

03.

Output:

Since 2024, three university teams at UC Santa Barbara and UC Irvine have advanced the detection research we began in 2023, across AI-generated music, vocals, and speech.

04.

Match:

We match AI generated songs back to the original recordings inside them. With UC Berkeley we now benchmark those methods so results can be tested, compared, and explained.

05.

Record:

We turned findings into a searchable record and are releasing it carefully, so artists, songwriters, and rights holders can look up their work and see what our research shows.

The Road Less Traveled

We believe it is our responsibility as artists to help define a responsible path for AI development.

‍

If we are going to say AI should not be developed a certain way, we also need to be able to point to the alternative: what should have been done and what was expected. Sound Ethics works with universities, researchers, artists, rights holders, and industry partners to turn those expectations into a practical roadmap for AI research and development.

‍

That includes how data is sourced and documented, how permission, attribution, and provenance are handled, and how responsible practices shape research and commercial use.

‍

Is this ok?

AI Rights

Fighting for Fair

Empowering artists to protect their work and know when it is used in AI

‍

The AI space is difficult to navigate and constantly evolving, with new methods, proposed frameworks, and a shifting legal landscape. We collaborate with legal experts and rights holders to improve transparency through independent research that supports informed decisions and helps protect artists and rights holders.

(01)

ATTRIBUTION

Music, Voice, NIL (Name, Image, Likeness)

What platforms can use your data?

Block: Data cannot be used

(02)

UNCOVER THE SONGS

Identify songs used in AI development

Every song leaves a trace

Connect findings to artists and rights holders

(03)

Detection & Provenance

Detect AI-generated music, vocals, and speech

Identify copyrighted material in Gen-AI outputs

Make AI-generated content more traceable

our ethos

WHAT MODELS ARE TRAINED ON SHOULDN’T BE A SECRET

We believe artists, songwriters, and rights holders deserve to know when their work trains AI. Our researchers study foundational AI music and audio machine-learning models, including which songs have been used in their development. Sound Ethics is carefully releasing its findings and preparing a public lookup tool that will make this history more visible and searchable.

Talk
with us

Talk with us

Whether you are from a university, part of the music industry, a data scientist, or involved with a government or non-profit organization, we welcome your interest in supporting our advocacy efforts.