Soma26 – Generative Neuro-Acoustic Engine

What is Soma26

Soma26 is a real-time audio application that uses brainwave entrainment to guide the listener into desired mental states — deep focus, creative flow, relaxation, or restorative sleep. Rather than playing pre-recorded tracks, it generates an infinite, evolving soundscape on the fly using psychoacoustic principles backed by neuroscience research.

The idea

Our brains naturally synchronize with rhythmic external stimuli — a phenomenon called neural entrainment. Soma26 exploits this by embedding precisely calibrated amplitude modulations into a layered audio texture. Unlike simple binaural beat generators, it applies modulation selectively to the 200 Hz–1 kHz frequency band (where entrainment is most effective and least fatiguing), while leaving bass and treble untouched. The result is a sound that feels like music, not a clinical tone.

Each mode targets a specific brainwave range:

• Focus — Beta waves (16 Hz) for sustained concentration
• Creative — Theta waves (6 Hz) for intuition and flow
• Relax — Alpha waves (10 Hz) for calm awareness
• Sleep — Delta waves (2.5 Hz) for deep rest

How it works

The audio pipeline stacks five functional layers: a noise blanket (pink/brown noise) that masks the environment, harmonic pads built from slow-evolving chord progressions, a spectral gate that applies the entrainment modulation to the mid-band only, a spatial processor that externalizes the sound (so it feels like it's coming from the room, not inside your head), and a subtle binaural carrier underneath. A probabilistic arranger — inspired by Brain.fm's approach — selects and transitions between musical motifs, varying density and intensity over time so the soundscape never loops or grows stale.

Chord progressions follow a harmonic grammar based on Martin Rohrmeier's linguistic model of tonal harmony, generating musically coherent sequences while staying within a low-cognitive-load pentatonic framework.

Technology

The core engine is written in Rust using the fundsp functional DSP library for real-time audio synthesis. It runs as a CLI tool, a WebAssembly module powering the web app, a native iOS framework (via UniFFI), and an Android/KMP library — all sharing the same Rust core. The web frontend is built with Next.js and the mobile app uses Kotlin Multiplatform with a native Swift layer on iOS.

Links

• Web app — Try it in your browser
• Landing page — Learn the science