Each speaker measured beautifully on its own. Left: ±2 dB from 65 Hz to 18 kHz. Right: ±2 dB from 65 Hz to 18 kHz. A matched pair that any manufacturer would be proud of. Then I measured both playing together at the listening position, and the response looked like it had been attacked with a serrated knife.

Peaks and nulls alternated every few hundred hertz, with some cancellations exceeding 15 dB. The coherence measurement dropped below 0.3 across most of the upper midrange. Two excellent speakers had combined to create something substantially worse than either one alone.

The problem was geometry. The two speakers were mounted 2.4 meters apart, aimed at a listening position that was slightly off-center. At 1 kHz, the wavelength is 34 centimeters. The path length difference between the two speakers at the measurement position was 17 centimeters — almost exactly half a wavelength at 1 kHz. Perfect cancellation. At 2 kHz, the path difference was a full wavelength — constructive addition. At 3 kHz, back to cancellation. A textbook comb filter, as predictable as sunrise.

Moving the measurement mic 30 centimeters to the left — to the exact center between both speakers — collapsed the path difference to zero. The comb filter vanished. The response was stunning. The room had one perfect seat, and nobody was sitting in it.

The Moral: Two sources means interference. Always measure the combined response at the actual listening position. SonaVyx's transfer function with coherence display reveals interference patterns instantly — so you can fix the geometry before the audience discovers the problem.