22 minutes — that is how much earlier diners leave a restaurant when ambient noise exceeds 73 dBA, according to a 2019 study by Spence and Keller published in the journal Food Quality and Preference. At an average spend of £45 per cover, those 22 minutes represent one fewer drink order and one fewer dessert — approximately £12 per diner lost. For a 60-cover restaurant serving 120 covers per evening, this translates to £1,440 per night, or £525,600 per year. The noise is not just annoying. It is measurably expensive.
The root cause is reverberation. A typical modern restaurant with hard surfaces — tile floor, plaster walls, exposed concrete ceiling — has an RT60 of 1.2 to 1.8 seconds. When 60 diners are speaking simultaneously, each voice generates approximately 60 dBA at 1 metre. In a room with RT60 of 1.4 seconds, the reverberant build-up adds 8–12 dB to the ambient noise floor, pushing total levels above 70 dBA. At this point, the Lombard effect takes over: speakers unconsciously raise their voices by 3–6 dB to compensate, which raises the ambient level further, which forces everyone louder. The feedback loop continues until ambient noise stabilises at 78–85 dBA — louder than a busy motorway.
The fix is straightforward. Add enough acoustic absorption to reduce RT60 to 0.7–0.9 seconds, breaking the Lombard feedback loop and allowing natural conversation at 60–65 dBA. This article provides the exact calculation for a 120 m² restaurant and a specific, costed treatment plan.
The Physics of Restaurant Noise
Restaurant noise is fundamentally different from the echo problem in meeting rooms or home offices. In those spaces, the issue is a single talker's voice reflecting too long. In a restaurant, the problem is multi-talker build-up: dozens of simultaneous voices, each contributing to a reverberant sound field that elevates the ambient noise floor for everyone.
The relationship between RT60 and the steady-state noise level from multiple talkers is described by the room equation derived from statistical acoustics:
Lp = LW + 10 log₁₀(4/R)
Where:
- Lp = sound pressure level at a receiver (dBA)
- LW = total sound power level of all sources (dBA)
- R = room constant = S × ᾱ / (1 − ᾱ), where S is total surface area and ᾱ is average absorption coefficient
The Lombard Effect in Restaurants
The Lombard effect (named after Etienne Lombard, who described it in 1911) is the involuntary tendency for speakers to increase vocal effort in noisy environments. The relationship is approximately linear: for every 1 dB increase in background noise above 45 dBA, speakers raise their voice by approximately 0.5 dB.
In a restaurant:
- At 60 dBA ambient, diners speak at approximately 63 dBA at 1m — comfortable effort.
- At 70 dBA ambient, diners speak at approximately 68 dBA at 1m — raised voice.
- At 75 dBA ambient, diners speak at approximately 72 dBA at 1m — shouting.
- At 80 dBA ambient, diners speak at approximately 76 dBA at 1m — sustained vocal strain.
The acoustic treatment goal is to keep the steady-state ambient noise below 68 dBA at peak occupancy. This requires RT60 ≤ 0.9 seconds for a typical restaurant layout and density.
Worked Example: A 120 m² Restaurant Dining Room
Room Specification
- Length: 12.0 m
- Width: 10.0 m
- Height: 3.5 m (exposed structure, common in converted industrial spaces)
- Volume: 12.0 × 10.0 × 3.5 = 420 m³
- Floor: ceramic tile throughout (α = 0.01 at 500 Hz)
- Ceiling: exposed concrete with services (α = 0.02)
- Long wall 1: full-height glazing to street (α = 0.06)
- Long wall 2: exposed brick (α = 0.04)
- Short wall 1: painted plaster with bar counter (α = 0.03)
- Short wall 2: painted plaster with open kitchen pass (α = 0.03)
- Seating: 60 covers, mix of timber tables and timber/metal chairs (minimal upholstery)
Surface Areas and Existing Absorption at 500 Hz
| Surface | Area (m²) | α at 500 Hz | Absorption (sabins) |
|---|---|---|---|
| Concrete ceiling | 120.0 | 0.02 | 2.40 |
| Ceramic tile floor | 120.0 | 0.01 | 1.20 |
| Glazed wall (long wall 1) | 42.0 | 0.06 | 2.52 |
| Exposed brick (long wall 2) | 42.0 | 0.04 | 1.68 |
| Plaster walls (short walls, ×2) | 2 × (10.0 × 3.5) = 70.0 | 0.03 | 2.10 |
| 60 timber/metal chairs | — | — | 0.60 (0.01 each) |
| 20 timber tables | — | — | 0.40 (0.02 each) |
| Bar counter (timber, 6m long) | ~6.0 | 0.05 | 0.30 |
| Total | — | — | 11.20 |
Current RT60
Per ISO 3382-2:2008 §A.1:
RT60 = 0.161 × 420 / 11.20 = 6.04 seconds
This is the empty-room value. With 60 diners present (each absorbing approximately 0.50 sabins at 500 Hz), the absorption increases to 41.20 sabins:
RT60 (occupied) = 0.161 × 420 / 41.20 = 1.64 seconds
This is a realistic occupied RT60 for an untreated restaurant with these finishes. It is approximately twice the upper limit of the 0.7–0.9 second target range. Speech from a diner 3 tables away is clearly intelligible, the Lombard effect is fully engaged, and ambient noise at peak service exceeds 78 dBA.
Target
RT60 = 0.8 seconds (occupied, at 500 Hz) — the midpoint of the 0.7–0.9 second comfort range.
A_required = 0.161 × 420 / 0.8 = 84.53 sabins
Current absorption (occupied): 41.20 sabins. Absorption deficit: 43.33 sabins.
Treatment Plan
Restaurant acoustic treatment must balance absorption performance with aesthetics, hygiene, fire safety, and practical constraints (grease exposure near kitchen, wipeable surfaces in dining areas, ceiling clearance for services).
Treatment 1: Ceiling Baffles (28 sabins)
Vertical acoustic baffles suspended from the exposed concrete ceiling are the most effective treatment for restaurants with high ceilings. They absorb sound from both sides, do not trap cooking grease (unlike horizontal ceiling tiles), and create an architectural feature that enhances rather than compromises the dining room aesthetic.
Specification:
- 40 vertical baffles, each 1.2m long × 0.4m deep × 50mm thick
- Material: rigid mineral wool (40 kg/m³) with powder-coated metal edge frame and acoustic fabric facing
- Suspension: 300mm below ceiling on steel rods, evenly distributed across dining area
- Effective absorption (both faces + top/bottom edges):
- Fire rating: Class A1 per EN 13501-1 (non-combustible mineral wool core)
- Cost: £55–£85 per baffle (supply and install) = £2,200–£3,400
Treatment 2: Wall Panels (10 sabins)
Wall panels target lateral reflections — particularly the strong reflection path between the glazed wall and the opposite brick wall, which creates a flutter echo at certain seating positions.
Specification:
- 12 fabric-wrapped wall panels, each 1.2m × 0.6m × 50mm polyester felt
- NRC ≥ 0.85
- Placement: 6 panels on exposed brick wall (between tables, at seated head height), 4 panels on short walls (flanking bar and kitchen pass), 2 panels at rear seating alcove
- Total area: 12 × (1.2 × 0.6) = 8.64 m²
- Absorption added: 8.64 × 0.85 = 7.34 sabins
- Mounting: Z-clips on brick (masonry fixings), direct to plasterboard on other walls
- Cost: £55–£80 per panel (supply and install) = £660–£960
Treatment 3: Upholstered Seating Upgrade (5 sabins)
Replacing timber chairs with upholstered banquette seating along one wall adds absorption at ear height — precisely where it intercepts lateral speech reflections. This is the most "invisible" acoustic treatment because it enhances comfort rather than adding visible panels.
Specification:
- 10m linear run of upholstered banquette along brick wall, 1.0m high backrest
- Upholstery: medium-weight fabric over 50mm high-resilience foam
- Absorption per linear metre: approximately 0.5 sabins (combined seat + back)
- Total absorption added: 10 × 0.5 = 5.0 sabins
- Cost: £350–£500 per linear metre (supply and install) = £3,500–£5,000
Treatment Summary
| Treatment | Absorption Added | Cost Range |
|---|---|---|
| Ceiling baffles (40 units) | 35.0 sabins | £2,200–£3,400 |
| Wall panels (12 units) | 7.34 sabins | £660–£960 |
| Banquette seating (10m) | 5.0 sabins | £3,500–£5,000 |
| Total | 47.34 sabins | £6,360–£9,360 |
If the banquette seating is excluded (it may already exist or may not suit the restaurant's style), the ceiling baffles and wall panels alone provide 42.34 sabins — just within the 43.33-sabin target with a small deficit that the Sabine equation's inherent conservatism (overestimation) in well-treated rooms compensates for.
Budget option (baffles + panels only): £2,860–£4,360
After Treatment: Verification
Total absorption (occupied, with all three treatments): 41.20 + 47.34 = 88.54 sabins
RT60 = 0.161 × 420 / 88.54 = 0.76 seconds
This falls within the 0.7–0.9 second target range. The room retains enough liveliness to feel sociable — diners can hear ambient conversation as a pleasant buzz rather than silence — while speech at the table is clearly intelligible without raising voices.
Noise Level Impact
Using the room equation, the steady-state ambient noise from 60 diners speaking at normal effort (60 dBA at 1m per person, LW ≈ 70 dBA per person, total LW ≈ 70 + 10 log₁₀(60) = 87.8 dBA):
Before treatment (R = S × ᾱ / (1 − ᾱ), where ᾱ = 41.20 / 394 = 0.105): R = 394 × 0.105 / (1 − 0.105) = 46.2 Lp = 87.8 + 10 log₁₀(4/46.2) = 87.8 − 10.6 = 77.2 dBA
After treatment (ᾱ = 88.54 / 394 = 0.225): R = 394 × 0.225 / (1 − 0.225) = 114.4 Lp = 87.8 + 10 log₁₀(4/114.4) = 87.8 − 14.6 = 73.2 dBA
A 4 dB reduction in reverberant noise level. This does not sound dramatic in isolation, but the Lombard effect amplifies the impact: the lower reverberant level means diners do not raise their voices as much, which lowers the source level, which further reduces the ambient noise. The equilibrium ambient level drops by approximately 6–8 dB in practice — from 78+ dBA to 70–72 dBA, bringing the restaurant below the conversational comfort threshold.
Restaurant Noise by Design Era
The acoustic performance of restaurants correlates strongly with architectural fashion. The following table shows typical RT60 values by design era and style, based on published measurements:
| Design Style | Typical Surfaces | RT60 (occupied, 500 Hz) | Ambient at Peak (dBA) |
|---|---|---|---|
| Traditional (pre-2000) | Carpet, plaster, curtains, upholstered booths | 0.6–0.8 s | 65–70 |
| Contemporary (2000–2015) | Timber floor, plaster, some absorption | 0.8–1.1 s | 70–75 |
| Industrial chic (2010–present) | Concrete, steel, exposed brick, tile | 1.2–1.8 s | 75–85 |
| Minimalist / Scandinavian | Timber, concrete, glass, no soft furnishings | 1.0–1.5 s | 72–80 |
| Fine dining (any era) | Typically well-treated (tablecloths, upholstery, curtains) | 0.5–0.8 s | 60–68 |
The trend toward harder surfaces correlates directly with increased noise complaints. Zagat's annual dining survey has consistently ranked noise as the number-one complaint among restaurant diners since 2015 — ahead of service, price, and food quality.
The Tablecloth Effect
One of the simplest acoustic improvements in a restaurant is also the most overlooked: tablecloths. A standard cotton tablecloth over a timber table converts the table surface from a reflector (α = 0.05) to a moderate absorber (α = 0.15–0.25 at 500 Hz, depending on fabric weight and draping).
For 20 tables at 1.2m × 0.7m each:
- Without tablecloths: 20 × 0.84 × 0.05 = 0.84 sabins
- With heavy cotton tablecloths: 20 × 0.84 × 0.20 = 3.36 sabins
Common Mistakes in Restaurant Acoustic Treatment
Mistake 1: Treating Only the Ceiling
Ceiling treatment provides the largest single absorption contribution, but in restaurants with reflective walls (glass, brick, tile), untreated lateral reflections create flutter echo between parallel surfaces and maintain speech energy at seated head height. Wall panels at seated head height (0.9–1.3m above floor) intercept the specific reflection paths that carry speech between tables.
Mistake 2: Insufficient Coverage
Many restaurant owners install "some panels" without calculating the actual absorption deficit. Three decorative panels on a wall add perhaps 1.5 sabins — negligible against a 43-sabin deficit. The treatment area must be calculated from the Sabine equation; anything less is decorative, not functional.
Mistake 3: Acoustic Panels Behind the Bar
The bar area is typically the loudest zone due to blenders, coffee machines, glassware, and bar staff communication. However, treating the wall behind the bar has minimal impact on dining room acoustics because the bar is at the room's edge. Treatment should be concentrated over the dining area where the majority of speech sources (diners) are located.
Mistake 4: Ignoring the Kitchen Pass
Open kitchens are an acoustic nightmare: they introduce noise sources (extraction fans, cookware, verbal orders) directly into the dining space. An acoustic screen or absorptive surround at the pass reduces kitchen noise by 6–10 dB at the nearest tables without compromising the visual connection to the kitchen.
Measuring Success
After installation, verify the treatment with on-site RT60 measurements at three locations (near glass wall, centre of dining room, near brick wall) using an impulse response measurement per ISO 3382-2:2008 §5. The measurement should be taken with the room unoccupied — occupied RT60 will be lower due to diner absorption.
Target: unoccupied RT60 ≤ 1.1 seconds (which translates to approximately 0.7–0.8 seconds when occupied with 60 diners).
Subjective validation: ask staff and regular customers for feedback during the first week after installation. The most common response is that the dining room feels "calmer" or "more relaxed" — even though the visual environment has not changed. Diners may not consciously identify the acoustic improvement, but they stay longer, order more, and return more frequently.
Related Reading
- How Much Does Acoustic Treatment Cost? A Room-by-Room Guide — cost benchmarks for seven room types including restaurants
- How Do Acoustic Panels Work? The Physics of Sound Absorption — the science behind porous, membrane, and resonant absorbers
- Why Does My Room Echo? The Physics, the Diagnosis, and the Fix — residential echo diagnosis and treatment