Attractor Labs
We build audio world models.
© 2026 Attractor Labs. All rights reserved.
Sound tells us information about the physical world that vision alone doesn’t. We hear an ambulance before we see it, and we can tell how big a room is from just its echoes.
We’re teaching AI to understand sound in space the way humans do. This means understanding how sound is produced, how it travels through space, and what it reveals about the world. We call this acoustic spatial intelligence.
Give it a world, and it should infer what it sounds like. Give it the sound of one, and it should know what’s happening and where.
SATIE, our first product, is an audio director for 3D worlds that does the first. LUDWIG, our perception model, does the second, and it learns from the worlds SATIE creates.
We’re a multidisciplinary team of researchers and engineers from Stanford, MIT, UC Berkeley, and Oxford, with expertise spanning machine learning, spatial computing, auditory neuroscience, and music technology. Our research has been featured at leading computing conferences including NeurIPS, CHI, UIST, and NIME.
We’re also artists and musicians, with backgrounds at Berklee College of Music and the Royal Conservatoire of Antwerp, and this shapes the question we care about and the technology we build.
We believe sound in space will be a fundamental part of physical AI and the future of media.