The audio-AI field is being built right now, and the choices made about data will decide whether artists ever trust it. I started Stylus in part to show that they should be able to, that high standards and doing right by musicians are not at odds with building great AI.
The audio I work with comes from real musicians and real sessions, not scraped from the web or synthetically faked. Human performances, from named sources.
I work directly with the artists behind the data and structure each partnership to be fair and transparent, something everyone involved can feel good about being part of.
I care about where data comes from and can speak to it. The sets you train on rest on a foundation you can defend, to your team, your users, and your lawyers.
I want a future where the technology and the people who make music both thrive. Every dataset I help build is a small argument that the two belong together.