Why 20 degrees? - Why the Best Crosstalk Cancellation Demands an Ambiodipole — and Why CTXMatrix Stays Pure

Ralph Glasgal’s one of his smaller setup with MBL speakers. Note the narrow speaker angle.

Crosstalk cancellation is not merely a technical trick. Its real purpose is to let the brain hear the interaural time differences (ITD) and interaural level differences (ILD) that were encoded in the recording, delivered as closely as possible to the way a natural, single-point sound source reaches us in real life.

In nature a sound arrives from one location. Both ears receive a coherent wavefront shaped by a single set of head-shadow and pinna cues. The brain integrates those consistent cues and localizes the source with ease. Conventional stereo does the opposite. Two widely spaced loudspeakers create two separate sources. Each ear hears both speakers. The resulting constant, artificial ITD and ILD from the loudspeakers themselves tell the brain: “There are additional sounds coming from the sides.” Comb filtering, conflicting pinna angles, and false head-shadows appear. Over time the brain tires of the inconsistency. Listeners sense something is not quite right and feel the urge to tweak tone, balance, or placement — because the cues are not those of real sound.

The only geometry that restores single-source-like delivery without adding further processing is the Ambiodipole: two loudspeakers placed close together, roughly 15–25° apart as seen from the listening position (around 20° is often ideal for large high-end systems). At this narrow frontal angle several physical and psychoacoustic advantages appear at once.

•  Path-length differences between the two speakers and the two ears become small. The delay and level of the cancellation signal needed to null the unwanted cross path therefore remain nearly constant across frequency. Simple polarity inversion plus a short delay works cleanly.

•  Head-shadow differences are minimized for the primary (especially central) signals. The cancellation parameters become far less dependent on individual head size and shape.

•  Central images arrive from roughly the correct frontal direction, so pinna cues remain appropriate and comb-filtering is pushed to higher, less audible frequencies.

•  Monaural spectral cues that arise when a source is off-centre are less disrupted. Wide-angle speakers produce strong, frequency-dependent diffraction and shadowing at each ear; these monaural localization features are extremely difficult to suppress with pure cancellation. Narrow placement keeps them more consistent with a natural frontal source.

CTXMatrix is deliberately pure cancellation — no equalisation, no complex HRTF modelling, no recursive EQ compensation. It relies on the acoustic geometry itself to keep the required correction simple and accurate. That is why the best results occur at the Ambiodipole angle. The same pure approach cannot be equally perfect at conventional 60° spacing. At wider angles the acoustic transfer functions change dramatically with frequency; head diffraction and path differences become highly frequency-dependent. Perfect broadband cancellation then requires additional equalisation to flatten the residual errors. Many audiophiles have historically disliked the sound of such EQ, and for good reason: any corrective filtering risks introducing its own artefacts or altering the very temporal and spectral cues we are trying to preserve.

I have tried to make CTXMatrix as usable as possible for listeners whose rooms force wider speaker angles. Yet my own system, built around rather large loudspeakers, does not allow thorough experimentation at those wider placements. The physics remains clear: pure cancellation reaches its highest fidelity only when the speakers themselves are close enough that the acoustic problem stays simple. In that sense the Ambiodipole is to crosstalk cancellation what a dedicated projection screen is to an 8K projector — the geometry that lets the technology reveal its full resolution instead of fighting the surface on which it is displayed.

Place the speakers at a true narrow frontal angle, engage pure cancellation, and the brain finally receives ITD and ILD the way nature intended: from a coherent, single-source-like field rather than from two competing point sources. The stage opens, the speakers disappear, and the constant urge to “tweak because something is artificial” falls away. That is the design goal of CTXMatrix.

ST