The principle — masking speech intelligibility
In an untreated open office, you can hear the conversation at the next desk clearly enough to understand every word. That's cognitively distracting — your brain processes the speech whether you want it to or not.
Sound masking adds a low, engineered background noise (typically 45–48 dBA of pink-noise-like signal) in the frequency range of speech. The intelligibility drops; you can hear that people are talking but can't make out words. The distraction goes away.
How masking systems are installed
Ceiling-mounted speakers connected to a masking signal generator. Speakers spaced every 15–20 feet across the floor plate at consistent volume. Modern systems calibrate automatically based on ambient noise levels.
The masking sound is barely perceptible on its own — it sounds like distant HVAC noise. But its effect on speech intelligibility is measurable and significant.
Masking + acoustic treatment — both, not either
Sound masking works best when reverberation is already controlled. In a reverberant open office, masking amplifies the reverberation and makes things worse.
The sequence: install acoustic treatment first (PET clouds, wall panels), measure RT60 into the target range, then add sound masking on top. Skip step one and step two makes things worse.