The Physics of Sound Reflection Explained

By Sarah Okonkwo · April 29, 2026

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

In February, SonusGearFlow was brought into a retrofit project at Riverside Arts Annex, a 240-seat black-box theater attached to a community arts center in Milwaukee, WI. The room was used for small concerts, spoken-word events, rehearsals, and occasional livestreamed panels. The team included one systems engineer (lead), one acoustical consultant, a project manager, and two installers from the venue’s preferred contractor. The venue’s technical director served as the client-side lead.

The “why” was simple and measurable: the room sounded inconsistent depending on seating location, and the venue was losing rentals. Musicians complained about “slapback” off the rear wall, presenters struggled with intelligibility, and the livestream feed (a stereo pair hung near FOH) sounded harsh compared to the in-room experience. The venue had already spent money on new loudspeakers the year before—QSC K12.2 on temporary stands—yet the improvements were limited because the primary issue wasn’t power or fidelity. It was reflection behavior: early reflections arriving with enough level and short enough delay to smear clarity, plus late reflections building an uneven decay.

We treated the job as an acoustics-first case study: use real measurements to identify reflection paths, then implement targeted controls with predictable outcomes. The work ran from initial survey to final commissioning in six weeks, with three on-site days for measurement and installation verification.

2) Challenges and requirements at the outset

The space was a rectangular room, 19.2 m long × 12.8 m wide × 5.6 m high (63 × 42 × 18 ft), with a flat painted drywall ceiling, polished concrete floor, and a mix of brick and drywall on the long walls. The stage was portable risers at one end. Seating was flexible, often 10–14 rows with a center aisle. The rear wall was a continuous painted cinder block surface with two steel doors—an ideal reflector.

Key constraints:

No permanent wall construction allowed (the venue needed to maintain multi-purpose use and had limited capital budget).
Aesthetic limits: the arts center wanted improvements that would not look like a “recording studio,” meaning minimal foam and no visible hanging baffles below lighting truss height.
Budget cap: $24,000 total for acoustic improvements and minor system changes.
Schedule: installation had to occur between rentals; only a 48-hour window was available for the main install.
Noise: the HVAC system produced 38–42 dBA in the mid-room—acceptable for most uses but sensitive for spoken word and recording. We were asked not to worsen it by adding rattling elements or obstructing airflow.

Performance requirements were defined up front so the results could be evaluated objectively:

Speech clarity: improve STI from ~0.45–0.50 to ≥ 0.60 in the central seating area using the existing reinforcement chain.
Reduce slapback: eliminate audible rear-wall echo for handclaps and amplified speech at typical levels (85–92 dBA at FOH).
Consistency: limit seat-to-seat variance in the 1–4 kHz region caused by strong early reflections.
Livestream sound: reduce harshness and comb filtering captured by the FOH stereo pair by controlling early reflections into that mic position.

3) Approach and methodology chosen

We framed the project around a practical explanation of reflection physics: sound reflects like light in many useful ways (angle of incidence equals angle of reflection), but the audible impact depends on time of arrival, level, and spectrum. Early reflections arriving within roughly 5–30 ms of the direct sound can smear intelligibility and produce comb filtering; later reflections contribute to perceived reverberance and can mask details if the decay is uneven.

Our methodology combined:

Impulse-response measurements to quantify arrival times and energy of reflection paths.
ETC analysis (energy-time curve) to locate problematic early reflections.
Geometric prediction (mirror-source method) to map which surfaces were responsible.
Targeted treatment where it mattered most: rear wall, first-reflection zones to audience and FOH mics, and ceiling contribution.
Minimal system changes to improve directivity and reduce excitation of reflective surfaces.

Tools and equipment included Room EQ Wizard (REW) on a Windows laptop, a Focusrite Scarlett 2i2 interface, a calibrated Earthworks M30 measurement mic, a NTi Minirator for spot checks, and a laser distance meter for geometry. For post-implementation verification, we repeated measurements at 12 audience locations plus FOH and stage positions.

4) Step-by-step execution narrative

Week 1: Site survey and baseline capture. We began with a walk-through during an empty room state and then during a rehearsal. The slapback complaint was immediately reproducible: a single clap at center seating produced a distinct echo, and amplified speech from stage returned from the rear wall with a delay that felt “late enough to be separate.”

Baseline measurements were taken with one K12.2 placed at stage-left as a reference source (we wanted a consistent, repeatable source position). The first major reflection peaks in the ETC occurred around 22–28 ms after the direct arrival at mid-room listening positions. Given the room length, this time window aligned closely with a rear-wall bounce (direct to rear wall, then back to listener). The level of that reflection was only 8–10 dB below direct in the 1–2 kHz range—high enough to be clearly audible.

We also saw pronounced early energy around 9–12 ms at FOH mic positions, consistent with ceiling reflections and lateral wall paths feeding the microphones. That aligned with the livestream harshness: the FOH pair was capturing a mix of direct plus early reflections, producing comb filtering that changed with mic height and small position shifts.

Week 2: Modeling and treatment plan. Using the mirror-source method, we mapped likely first-reflection points on the long walls and ceiling for typical source (stage) to listener positions. We prioritized surfaces that contributed strong reflections in the 1–4 kHz band, where intelligibility lives and where comb filtering is most audible.

We proposed three physical interventions:

Rear-wall absorption to kill slapback without deadening the entire room.
Selective ceiling absorption above the first third of the audience area and FOH region to reduce early reflections into listeners and microphones.
Limited diffusion on rear wall flanks to keep the room from becoming unnaturally dry and to scatter any residual energy.

Week 3–4: Procurement and pre-build. We chose GIK Acoustics 244 bass trap/absorber style panels (4-inch mineral wool) for broadband performance. For ceiling, we used 2-inch fiberglass panels (OC 703 equivalent) in low-profile metal frames with white acoustically transparent fabric. We selected these because they performed predictably, were fire rated, and could be installed quickly.

Week 5: Installation. The contractor installed rear-wall panels and ceiling treatment in a single overnight shift plus a half-day punch list. The install details mattered:

Rear wall: 18 panels of 24 × 48 × 4 in (total coverage ~144 sq ft). We mounted them with a 2-inch air gap behind each panel to increase low-mid absorption efficiency.
Rear wall diffusion: four 2D quadratic residue diffusers (24 × 24 in) mounted at ear-height zones near the corners—not to “fix reverb,” but to break up specular returns and reduce the chance of a single strong reflected ray.
Ceiling: 10 panels of 24 × 48 × 2 in arranged in two rows above the audience front half and FOH line. Panels were mounted above lighting sightlines, with aircraft cable and safety ties, maintaining minimum 18 inches clearance from HVAC diffusers.

Week 6: Commissioning, tuning, and verification. We repeated measurements at 12 seats (three rows × four lateral positions), plus two stage positions and FOH. We also conducted a “real use” test: a lav mic for speech, a handheld dynamic (SM58), and a small acoustic trio with the same PA and typical gain structure.

5) Technical decisions and trade-offs made

Absorption vs. diffusion on the rear wall. The rear wall was the primary slapback source, so absorption was non-negotiable. The trade-off was avoiding an overly dead back-of-room that could feel unnatural for unamplified performances. We handled this by using broadband absorption for the central rear wall where reflection energy was strongest, and diffusion on the flanks. Diffusion doesn’t remove energy; it redistributes it in time and angle, lowering the chance of a single dominant echo.

Why 4-inch panels with an air gap. A 4-inch absorber with an air gap improves absorption down into the low mids (around 125–250 Hz), where “boom” and muddiness can build up. Slapback is often blamed on mids/highs, but the perceived “size” of the echo can include low-mid energy. The air gap was a cost-effective way to increase performance without thicker panels.

Ceiling treatment location. We did not blanket the entire ceiling; that would have been expensive, visually intrusive, and potentially too absorptive. The ETC showed strong early ceiling returns into FOH and front/mid seating. Treating targeted zones reduced early reflection energy where it mattered most while preserving some liveliness in the rear half of the room.

Small PA adjustment instead of replacement. Although the venue used portable K12.2 boxes, replacing them wasn’t in scope. We did, however, change deployment: we moved from wide spacing on stands to a tighter L/R with a slight downward tilt and reduced splay to keep energy off the side walls. We also standardized speaker height to 2.3 m (7.5 ft) horn height and applied modest HF shelving (-1.5 dB above 6 kHz) to reduce perceived harshness once reflections were controlled.

6) Results and outcomes with specific details

The improvements were measurable and audible.

Rear-wall reflection reduction: The ETC peak at 22–28 ms dropped by 11–14 dB across mid-room seats in the 1–2 kHz octave bands. Subjectively, the handclap echo changed from a distinct repeat to a short, non-distracting tail.
Speech intelligibility: STI improved from 0.47 average (center seating) to 0.63 average with the same mic and loudspeaker system. The largest gains were in the mid and rear seating where slapback previously masked consonants.
RT60 / decay behavior: Measured T20/T30 estimates (room occupied estimate adjusted via seating absorption assumptions) showed midband decay reduced from roughly 1.35 s to 0.95 s at 1 kHz. More importantly, the decay became smoother; the previous “hang” in the 500 Hz band was reduced by about 0.25 s.
Livestream capture: The FOH stereo pair (small-diaphragm condensers) produced less comb filtering; frequency response variance with a 10 cm mic move reduced visibly in REW. The technical director reported needing less corrective EQ and de-essing on recorded panels.
Gain before feedback: With a lav mic and handheld, the venue gained about 3–4 dB additional GBF in typical configurations. This wasn’t due to “magic panels,” but because fewer early reflections were re-entering open mics and re-exciting the room.

Budget and timeline landed as planned: $22,600 all-in including hardware, labor, and a second measurement day for verification.

7) Lessons learned and what could be done differently

Don’t guess the reflection path—measure it. The venue initially assumed the side walls were the main issue because they were hard surfaces and close to seating. Measurements showed the rear wall was the dominant slapback source due to room length and its uninterrupted reflective area. A small time-of-flight calculation (distance and speed of sound) lined up exactly with the 22–28 ms ETC peaks, reinforcing the physics with data.

Early reflections affect microphones as much as listeners. The livestream problem wasn’t solved by changing microphones. It improved when we reduced early ceiling and wall returns into the FOH position. For rooms doing both live reinforcement and recording, treat mic positions like critical listening positions.

Partial coverage can outperform blanket treatment when guided by ETC. We did not have the budget to treat everything. By focusing on surfaces responsible for the strongest early energy, we achieved most of the benefit. If budget allowed, the next increment would be additional ceiling panels over the mid-room to further stabilize clarity for music-heavy events.

What we would do differently: We would push earlier for a modest change in the PA approach—specifically, deploying a more controlled-pattern main speaker (even a compact constant-directivity point source) to reduce wall excitation further. The existing K12.2s worked, but their coverage is broad; room treatment had to carry more of the load.

8) Takeaways applicable to other projects

Reflection physics is operational, not theoretical. If you can identify the surface, the path length, and the arrival time, you can predict what the audience hears and what the microphones capture. ETC plots turn “it sounds echoey” into actionable targets.
Prioritize reflections by time and level. A reflection arriving 25 ms late and only 8 dB down is more damaging to intelligibility than a later, quieter wash. Fix the strongest early offenders first: rear wall, ceiling above FOH, and first-reflection zones to critical seats.
Use absorption where you need to remove energy; use diffusion where you need to keep life while avoiding specular returns. Rear walls often benefit from a hybrid approach: absorption to kill slapback, diffusion to prevent a dead acoustic and to reduce localized hot spots.
Small geometric changes in loudspeaker placement can reduce reflection problems. Before buying new gear, aim loudspeakers to minimize energy aimed at large reflective planes. A few degrees of tilt and a tighter splay can reduce early reflection level significantly.
Define success metrics before you begin. STI targets, ETC peak reductions, and documented mic capture improvements make the project legible to stakeholders and reduce subjective debates. Project managers benefit from these metrics because they align budget with measurable outcomes.

This project reinforced a consistent pattern: many “PA problems” are reflection problems first. When you treat the right surfaces for the right reasons—based on arrival time, energy, and geometry—you can achieve clearer speech, more stable music reinforcement, and better recordings without rebuilding the room.