Acoustic Modal Resonance in Open-Plan Offices

By Priya Nair · May 4, 2026

1. Project overview: what, where, who, and why

The project took place on the 8th floor of a renovated 1990s commercial building in Austin, Texas. A SaaS company had expanded into a 9,800 sq ft open-plan office with 84 desks, a “town hall” zone for all-hands meetings, and a lounge area with a small AV setup for presentations and hybrid calls. The space was visually impressive—polished concrete floor, exposed deck ceiling at 10 ft 6 in, glass-fronted huddle rooms on one side, and a long plasterboard wall on the other.

Within three weeks of move-in, operations reported a pattern: speech on video calls sounded “boomy,” and in-person meetings in the town hall were fatiguing. Employees described a strange “pressure” in the room when a male voice spoke near the lounge. The AV integrator initially suspected microphone EQ or echo cancellation tuning. The IT team had already swapped USB speakerphones and tried different conferencing platforms. Nothing stuck.

SonusGearFlow was brought in under a joint scope with the client’s facilities project manager and the existing AV integrator. Our mandate was not to redesign the office, but to identify the root cause, document it in measurable terms, and deliver a corrective package that could be implemented during business hours with minimal disruption.

2. Challenges and requirements at the outset

The client set practical constraints that shaped every technical decision:

No major construction: no new walls, no ceiling grid. The exposed ceiling was a design feature.
Operational continuity: work continued 9 a.m.–5 p.m. Treatments had to be installed in evenings and weekends, and any testing needed to be fast and repeatable.
Speech intelligibility first: the space wasn’t for music playback; the priority was better clarity for all-hands and hybrid calls.
Budget cap: $28,000 total, including materials, install labor, and measurement time.
Aesthetic limits: visible acoustic treatment had to match the office palette—charcoal, warm gray, and wood accents. “Studio foam” was explicitly rejected.

From our first walk-through, the acoustic signature was obvious: sharp flutter between the glass line and the long gypsum wall, and a low-frequency buildup in the lounge corner near the AV display. The floor plan included a nearly rectangular “town hall” zone measuring roughly 38 ft (L) × 24 ft (W) × 10.5 ft (H). That geometry is favorable for strong room modes, particularly when boundaries are hard and parallel.

3. Approach and methodology chosen

We treated this like a forensic investigation: confirm the complaint, quantify it, model likely modes, then choose a mitigation that fit the constraints.

Our toolkit and workflow were selected for speed and repeatability:

Measurement software: Room EQ Wizard (REW) for swept sine and decay metrics; SMAART v9 for quick transfer-function checks during live mic tests.
Measurement microphones: MiniDSP UMIK-1 for quick distributed sampling, and an Earthworks M30 for verification and consistency when comparing before/after data (calibrated file loaded).
Interface and playback: RME Babyface Pro FS feeding a QSC CP12 as a reference loudspeaker (placed at typical talker position), level-calibrated to 76 dBA at 1 m for test sweeps.
Noise logging: a handheld Class 2 SPL meter (Extech) for background measurements at different times of day.

We scheduled a two-phase evaluation: a four-hour after-hours baseline measurement session (to avoid occupancy noise), followed by two short daytime validation sessions where we could test conferencing behavior and subjective perception in real use.

4. Step-by-step execution narrative

Day 1: Baseline survey (after hours)
We started with a structured listening walk using pink noise and spoken word playback. The “boom” localized near the lounge—about 12 ft from the glass huddle room frontage and 6 ft from the gypsum wall. The effect wasn’t subtle: certain voice notes around the 80–120 Hz region would hang in the room, while nearby positions sounded more neutral.

We then ran REW sweeps at eight mic positions: four in the town hall seating area, two at standing presentation positions, and two in the lounge. At each position we captured frequency response and waterfall/decay plots.

Findings were consistent across the space, but worst in the lounge corner:

A strong peak centered at 94 Hz (+9 to +12 dB depending on position)
Secondary peaks around 63 Hz and 125 Hz
Low-frequency decay times (T60 proxy) exceeding 0.75 s at 94 Hz, while midband decay was closer to 0.45 s

We also measured a pronounced flutter echo between the glass and gypsum surfaces. Clapping tests were supported by impulse response captures showing distinct early reflections in the 12–25 ms window, aligning with the wall-to-wall distance.

Day 2: Quick modeling and hypothesis
Using the approximate room dimensions (38 ft × 24 ft × 10.5 ft), we computed axial mode estimates. The width mode for 24 ft predicts around 23.5 Hz (too low to be the main issue), but the height mode around 10.5 ft predicts about 53.8 Hz. The length mode around 38 ft predicts about 14.9 Hz. Those alone didn’t explain the 94 Hz dominance, but the combination of tangential/oblique modes and the boundary conditions (glass wall stiffness vs gypsum compliance, plus a large display wall) commonly creates pronounced modal hot spots in that region.

The 94 Hz peak also lined up with a practical observation: male voice fundamentals and first harmonics can strongly excite modes between 80–150 Hz, creating “chestiness” that microphones and echo cancellers struggle to manage. In conferencing, this manifests as aggressive AEC behavior and an unnatural pumping sound as the algorithm fights the room.

Day 3–7: Treatment plan, mockups, and approvals
We proposed a three-part mitigation package:

Broadband absorption overhead to reduce overall reverberant field and early reflections without altering the exposed ceiling look.
Targeted low-frequency control focused on the 80–125 Hz band using practical “thick” absorbers in corners and along boundary junctions where pressure maxima occur.
AV tuning adjustments after acoustic changes, not before, to avoid chasing EQ as a band-aid for modal behavior.

For aesthetics and install speed, we selected 2 in and 4 in PET-felt baffles and fabric-wrapped mineral wool panels in standard sizes. The client approved samples in charcoal fabric and a warm gray PET that matched existing desk dividers.

Week 2: Installation (evenings)
Installation occurred over three evenings and a Saturday morning. We coordinated with facilities so that desk areas remained usable, and we kept ladder work confined to after-hours.

5. Technical decisions and trade-offs made

Decision 1: Ceiling baffles vs full ceiling clouds
A full ceiling cloud would have been acoustically efficient but visually heavy. Instead, we specified 36 PET baffles (48 in × 12 in × 2 in) hung vertically above the town hall and main desk aisles. The vertical baffle approach increased effective absorption by exposing both sides of each panel and maintained the “open” ceiling aesthetic. Trade-off: baffles have less low-frequency absorption than thicker horizontal clouds, so we paired them with targeted LF measures.

Decision 2: Low-frequency control via thickness and placement, not “bass traps everywhere”
With a $28k cap, we couldn’t blanket the perimeter with 6–8 in traps. We prioritized positions with the highest modal pressure:

Two 4 in thick fabric-wrapped panels (24 in × 72 in) straddling the lounge corner (air gap ~2 in behind).
Four 4 in panels placed at wall-ceiling junction points above the town hall screen wall, mounted as “soffit-like” absorbers using brackets and a 3 in air gap.
Six 2 in panels (24 in × 48 in) on the long gypsum wall at seated head height to reduce early reflections and flutter with the glass.

Trade-off: these measures would not “flatten” the room like a studio. The goal was to reduce the severity of the 94 Hz ringing and improve speech clarity, not achieve a perfectly neutral response at every seat.

Decision 3: Avoid narrowband resonators for schedule and risk reasons
Helmholtz or membrane traps tuned to ~94 Hz could have been effective, but they require precise build and placement, and errors can waste budget quickly. Given the time constraints and the need for predictable results, we stayed with thick porous absorption and strategic placement. The client also preferred off-the-shelf products for procurement and future expansion.

Decision 4: AV EQ only after acoustic work
The integrator had attempted to EQ the speaker and tweak mic processing. We asked them to revert to a known baseline (flat system EQ, standard AEC settings) before post-treatment testing. The trade-off was short-term inconvenience, but it prevented us from “baking in” compensations that would be wrong once the room changed.

6. Results and outcomes with specific details

Measurement deltas (before vs after)
We re-measured the same eight mic positions with identical loudspeaker placement and level. The most meaningful improvements were in the low-frequency decay and the 94 Hz peak magnitude:

94 Hz peak reduction: from +9 to +12 dB down to +3 to +6 dB depending on location (largest improvement in the lounge corner).
Decay at 94 Hz: reduced from ~0.75–0.85 s down to ~0.45–0.55 s (waterfall plots showed a clearly shortened tail).
Early reflections: impulse response energy in the 12–25 ms window dropped by ~4–6 dB in the town hall measurement positions, aligning with reduced flutter perception.

Conferencing performance
The client’s primary pain point was hybrid calls. We conducted a controlled test using the installed conferencing system (ceiling mics and DSP provided by the integrator) plus a “worst-case” scenario using a USB speakerphone placed on a lounge table.

Post-treatment, AEC behavior improved in a way that matched the measurements: far-end participants reported fewer instances of “hollow” or “pumping” sound when someone spoke loudly near the lounge. Locally, talkers perceived less “pressure” on certain notes.

Operational impact
Installation was completed in 10 calendar days from baseline measurement to final verification, with only one Saturday morning requiring partial access restrictions. Total spend came in at $26,400, broken down roughly as:

$12,900 materials (baffles, panels, hardware)
$8,100 installation labor (including lift rental for one day)
$5,400 measurement, documentation, and coordination

Subjective outcomes
The office didn’t become silent—nor was that the goal in an open plan. But speech became less tiring in the town hall. The most telling feedback came from the project manager, who noted that all-hands Q&A no longer required “resetting” the room every time someone moved from the seating area to the lounge.

7. Lessons learned and what could be done differently

Lesson 1: Open-plan offices still behave like rooms
It’s tempting to assume an open-plan space is too irregular to have strong modal behavior. In practice, a “zone” within an open plan can be modal if it has consistent boundaries—glass on one side, gypsum on the other, hard floor, and a consistent ceiling height. The town hall zone functioned like a rectangular room embedded inside a larger space.

Lesson 2: Don’t let conferencing DSP mask acoustic root causes
AEC and noise reduction can hide problems until someone speaks at just the wrong pitch and level. The 94 Hz mode wasn’t solved by EQ because it was primarily a decay problem, not just a steady-state magnitude problem. If we had continued down the “tune the DSP” path, we would have traded one artifact for another.

Lesson 3: More measurement positions beats one “perfect” measurement
The modal hot spot was location-dependent. If we had measured at a single reference point, we could have missed how severe the lounge corner was. Eight positions was a manageable compromise: enough to see patterns, not so many that we lost the schedule.

What we would do differently
If we had been involved during design, we would have recommended two preemptive changes that cost less than retrofit:

Add absorption as part of the architectural ceiling plan—either discreet clouds or higher-NRC materials integrated into lighting lines.
Break up the largest parallel surfaces early (e.g., partial-height slatted wood with absorption backing on the long gypsum wall). It would have reduced flutter and improved aesthetics without looking like an “acoustic fix.”

8. Takeaways applicable to other projects

Identify whether the complaint is amplitude or decay. A modal peak that rings will sabotage speech and AEC. Treat decay with absorption and placement before reaching for EQ.
Use geometry to predict where problems will be. Corners and boundary junctions are high-pressure zones; they’re where thick absorption earns its keep, even in offices.
Choose treatments that fit operational reality. Vertical baffles and wall panels are often easier to install after hours than large ceiling clouds or construction-heavy solutions.
Document before/after with repeatable conditions. Same speaker, same mic positions, same level, same time window. The credibility of the fix matters when budgets are under scrutiny.
Coordinate with AV integrators instead of working around them. Reset DSP to baseline, implement acoustic changes, then retune. This sequence prevents circular troubleshooting.
Plan for “zone acoustics” in open plans. If a town hall zone is effectively a room, treat it like one—especially if it includes a presentation wall, glass frontage, and hard floors.

In this case, the solution wasn’t exotic. It was disciplined: confirm the mode, reduce the energy feeding it, shorten its decay, then retune the system around a healthier acoustic baseline. For project managers, the key was sequencing and constraints management. For audio engineers, the key was resisting the urge to solve a time-domain problem with frequency-domain tools alone.