Speech intelligibility, meeting privacy and open-space quiet.
When you have to ask people to repeat themselves in a meeting, and your voice sounds «barrel-like» on a video call — the problem is almost never the headset or camera. The problem is that speech reflects several times off glass, concrete, and drywall before reaching the listener’s ear or the microphone. These reflections overlay the direct sound and smear the consonants — and it’s the consonants that carry intelligibility.
The key «boominess» metric is the reverberation time RT60. Per the RoomPerfect methodology, acoustically treated spaces have an RT60 around 0.3–0.6 s, while boomy rooms with hard surfaces reach 1.5–5 s. The longer the reverberation «tail,» the more each next word lands on top of the previous one — and the harder it is to hear.
Intuition says we hear a speaker or a talker directly. In reality — no. Harman research (Toole/Olive) shows that beyond the critical distance in an ordinary space, what reaches the ear is mostly the reflected field, not direct sound. In perceptual weighting, direct sound accounts for only about 12–14%, while early reflections are about 44% and the late reverberant tail another 44%.
The direct conclusion for an office: the quality of what an employee or a caller hears is ~86% determined by how the space handles reflections, and only to a small degree by the speech source itself. So in a meeting room there’s no point starting with an expensive conferencing system. You have to start with the room.
In an office and a meeting room there’s a band where the space ruins the sound physically and predictably — below the transition frequency of ~300 Hz. Here the room modes and boundary effects rule. Per Harman research (Toole/Olive), source placement and proximity to walls alone make the response in this region vary by more than 18 dB, and between different rooms the differences reach 25 dB below 100 Hz.
This is the very «droning» lower midrange that makes a male voice in a meeting room sound heavy and unintelligible. The scale of the error — tens of decibels — is an order of magnitude larger than the difference between a good and a bad microphone or speaker. So the solution lies in the geometry and treatment of the space, not in the choice of equipment.
We work by the logic of the RoomPerfect methodology: first separate the source signature from the room acoustics, then build a model of the room from several measurements and correct exactly the room’s damage rather than imposing an abstract «flat» curve. As Peter Lyngdorf puts it: «It’s a mistake to target a flat frequency response unless you live in an anechoic chamber.»
For speech intelligibility this matters: the goal is not to «flatten a graph on paper» but to reduce early reflections and the reverberant tail so that consonants don’t smear — at every workstation and every chair at the conference table.
The microphone catches not only your direct voice but also its reflections off glass, concrete, and drywall. Per Harman research (Toole/Olive), in an ordinary space beyond the critical distance the receiver mostly gets the reflected field — direct sound is only ~12–14%. The longer the reverberation (RT60 of boomy rooms reaches 1.5–5 s versus 0.3–0.6 s in treated spaces), the stronger the echo in the mic. The solution is to damp the early reflections and shorten RT60, not to change the headset.
Equipment should be chosen after the room, not instead of it. The perceptual weighting from Harman research (Toole/Olive) — about 12–14% direct sound, ~44% early reflections, ~44% late tail — means speech quality in a space is almost 86% determined by the room itself. Below ~300 Hz, placement and boundaries give a spread of more than 18 dB — an order of magnitude larger than the difference between microphone models. The room first, then the equipment.
A 3D scan gives precise geometry: volume, the areas of reflective surfaces, a modal map. By the logic of the RoomPerfect methodology we separate the source signature from the room acoustics and correct exactly the room’s damage rather than imposing a flat curve. As Peter Lyngdorf says: «It’s a mistake to target a flat frequency response unless you live in an anechoic chamber.» The goal is a short RT60 of 0.3–0.6 s and clean consonants at every seat.
Most open-space and meeting-room tasks are solved with surface-mounted elements — calculated absorption and diffusion, without dismantling the finish. We integrate the acoustic elements into the interior by color and texture together with the designer. Control of source directivity (a verified precedent — Kii Audio: rear radiation at 200 Hz is 10–15 dB below the front) further reduces reflections without visible wall treatment.
Describe the space and the task — we’ll reply within 2 hours with a preliminary estimate