Skip to main content
Back

Bang & Olufsen Researchers Fix Room Audio by Adding a Hidden Second Speaker

A second, delayed loudspeaker adds frequency-selective energy to the reverberant field, correcting both timbre and spatial perception without listeners noticing it.

Pro

  • New room correction method uses a delayed, decorrelated second speaker to fill reverberant field, correcting spatial and spectral errors. Paper
  • Traditional inverse filters leave the Direct-to-Reverberant Ratio uncorrected because they scale direct and reverb equally.
  • A 10 ms delay activates the precedence effect so listeners localize only the primary speaker, not the support.
  • Filter is capped at 10 dB (low freq) / 6 dB (high freq) above the primary to preserve precedence.
  • Preferred over uncompensated stereo (p<0.05); statistically tied with a commercial room-correction algorithm.
  • Technically weaker than inverse filtering (4.5 dB vs 1.1 dB spectral deviation); no code released.

Room correction has been solving half the problem for decades. The standard fix for a bad-sounding listening room is to measure the loudspeaker's response at the listening position and apply an inverse filter to flatten the spectrum. That handles timbre. It does nothing for the way a room smears distance and spatial cues across frequencies. A new paper in the Journal of the Acoustical Society of America from researchers at Oldenburg, Aalborg, and Bang & Olufsen proposes a different approach: instead of filtering the primary speaker, add a second one that fills in the reverberant field on purpose.

Why inverse filters hit a wall

The core issue is the Direct-to-Reverberant Ratio (DRR). Humans rely on DRR for source distance perception, and variations across frequencies or position can shift the spatial image of the source. A conventional two-way speaker radiates low frequencies almost omnidirectionally while beaming highs on-axis, so more bass energy hits the walls than treble. Combined with a room's uneven absorption, DRR swings wildly with frequency, and the perceived distance of the source shifts along with it.

Traditional room compensation applies a filter w(ω) to the speaker's feed. While this controls the spectral energy distribution at the listening position, DRR is left uncontrolled unless full dereverberation is achieved. The filter scales direct and reverberant sound by the same amount, so their ratio never changes. Inverse filters also tend to introduce pre-ringing artifacts, audible resonances from inverted notches, and corrections that only work at one measurement position.

Hiding a second speaker in plain hearing

The proposed method sidesteps all of that. Adding a supporting loudspeaker alongside the primary one lets you inject energy into the room in a frequency-selective way, compensating for spectral notches in the primary source's output. Delaying the supporting source activates the precedence effect, so only the primary loudspeaker is spatially perceived as a source.