BSE

THE PROBLEM

Binaural is stuck with other people's ears

Binaural audio promises the ultimate spatial experience: full 3D positioning through ordinary headphones. The mechanism has been understood for over a century — your brain localizes sound through interaural time differences, level differences and the spectral fingerprint of your outer ears. Reproduce those cues accurately and the brain hears sound in space around you.

The traditional approach convolves audio with HRTFs — impulse responses measured on a dummy head or a small sample of subjects. But your head and ears are unique: the spectral cues that carry elevation vary so much between individuals that what sounds 'above' for one person sounds 'in front' for another through the same generic HRTF.

Convolution also carries a structural bill: continuous FFTs per source, CPU that scales linearly with source count, 10–50 ms of block latency that breaks audio-visual sync in interactive use, and embedded platforms that simply cannot run it. The industry treats these limits as inherent to binaural. They are not — they are inherent to convolution.

HOW IT WORKS

How it works

UNDER THE HOOD

Going deeper

BSE abandons HRTF convolution entirely and models each mechanism of spatial hearing explicitly, drawing on decades of psychoacoustic research: interaural time differences dominate localization below 1.5 kHz; level differences take over above, as the head shadows sound; the pinna's frequency-dependent reflections encode elevation; torso, shoulder and ground reflections contribute front-back discrimination and height cues. Each phenomenon becomes a parametric filter — no FFT, no convolution, no block processing.

The consequence is structural: a few samples of latency at any sample rate, a per-source cost so low that a single DSP core renders dozens of sources, and platforms convolution can never reach — Bluetooth headphones with onboard DSP, automotive head units, mobile without battery drain.

And because the processing is parametric rather than data-driven, personalization is real. Head width sets the interaural time differences your brain actually expects; head height adjusts torso and ground reflection geometry. These are not abstract numbers — they are dimensions you measure with a tape, and entering them improves spatial accuracy immediately, especially for elevation, where individual variation matters most. The rendering adapts to you, instead of forcing you to adapt to a dummy head.

Paired with HSR, BSE completes a stereo-to-binaural pipeline: HSR extracts the spatial content of a stereo mix as discrete components — the center vocal at 0°, the left guitar at −30°, the diffuse ambience across the rear hemisphere — and BSE renders each with full spatial cues. Streaming music gains genuine immersive presentation, in real time, without touching the source material.

AT A GLANCE

At a glance

PARAMETERS

Hands on the algorithm

POSITIONING

Compared to the alternatives

APPLICATIONS

Where it fits