Restaurant Acoustic Design — Why 80% of New Restaurants Are Too Loud | AcousPlan

# Restaurant Acoustics: Why 80% of New Restaurants Are Too Loud

In 2024, the food criticism platform Eater published an analysis of restaurant reviews across major US cities. Noise was the second most common complaint after food quality — ahead of service, price, and wait times. The same pattern appears in UK and Australian restaurant data. Restaurants that close within two years cite "ambiance problems" in a significant portion of post-mortem analyses, and when that data is disaggregated, noise is the dominant ambiance complaint.

The restaurants are not too loud because of bad equipment or loud customers. They are too loud because 80% of new restaurant spaces are designed with RT60 values that guarantee acoustic disaster at full occupancy. The acoustic behaviour of a restaurant under full service load can be predicted at the design stage with 15 minutes of calculation. It almost never is.

Here is what is happening, why interior designers get this wrong every time, and how to specify a restaurant that is energetic but not exhausting.

The Lombard Effect: How Rooms Amplify Themselves

The fundamental problem with restaurant noise is not the noise source — it is a feedback loop called the Lombard effect. When background noise rises, people unconsciously raise their voices to maintain intelligibility across the table. This increases the background noise level for everyone else in the room, who then raise their voices further. The room's reverberation time determines how quickly the cycle stabilises and at what noise level.

In a well-designed restaurant (RT60 ≈ 0.7 s at 500 Hz), the Lombard cycle stabilises at approximately 68–72 dBA. In a poorly designed restaurant (RT60 ≈ 1.5–2.0 s), the cycle stabilises at 78–85 dBA. The additional 10 dB is not because more noise is being generated per person — it is because the room is recycling each voice back into itself for 2–3 seconds longer before decay, and each recycle contributes to the sustained level.

The maths: in a diffuse reverberant field, sustained noise level Lp from multiple independent sources is approximated by:

Lp = Lw_per_person + 10 log₁₀(N) + 10 log₁₀(4/R)

where N is occupant count and R is room constant (m²):

R = A / (1 - ᾱ)

with A = total absorption in m² and ᾱ = mean absorption coefficient.

For a 150-seat restaurant, 300 m², 4m ceiling height = 1200 m³. Assume each person + conversation contributes approximately Lw = 55 dB (typical speech + background movement):

Case 1: Acoustic treatment included (RT60 ≈ 0.8 s)

A = 0.161 × V / RT60 = 0.161 × 1200 / 0.8 = 241.5 m²
ᾱ ≈ A / S_total = 241.5 / (300+300+4×√(300)×4) ≈ 0.45
R = 241.5 / (1 - 0.45) = 439 m²

Lp = 55 + 10 log₁₀(150) + 10 log₁₀(4/439)
   = 55 + 21.8 + 10 log₁₀(0.0091)
   = 55 + 21.8 + (-20.4)
   = 56.4 dB (at a receiver in reverberant field)

Add direct field contribution from nearby conversations and actual levels at seat ≈ 68–72 dBA.

Case 2: No acoustic treatment (RT60 ≈ 1.8 s)

A = 0.161 × 1200 / 1.8 = 107.3 m²
ᾱ ≈ 0.20
R = 107.3 / (1 - 0.20) = 134 m²

Lp = 55 + 21.8 + 10 log₁₀(4/134)
   = 55 + 21.8 + 10 log₁₀(0.0299)
   = 55 + 21.8 + (-15.2)
   = 61.6 dB (reverberant field)

Add Lombard effect uplift of approximately 6 dB (voices raised in response to elevated background) and actual seat levels ≈ 78–82 dBA. The difference between the two cases — 10+ dB — is entirely attributable to RT60.

Why Interior Designers Always Get This Wrong

The aesthetic preferences of contemporary hospitality design are directly opposed to acoustic performance:

Every trend in contemporary restaurant design — the stripped-back industrial aesthetic, the exposed services, the terrazzo floors — is an acoustic disaster. Individually, each choice seems reasonable. Collectively, they produce a room where the total absorption at 500 Hz might be:

A_total = 300 × 0.02 (concrete floor)
        + 300 × 0.02 (concrete ceiling)
        + 280 × 0.04 (glass + brick walls)
        + 150 × 0.05 (timber chair backs, upholstery)
        = 6.0 + 6.0 + 11.2 + 7.5
        = 30.7 m²

For RT60:

RT60 = 0.161 × 1200 / 30.7 = 6.3 s

That is an empty room. With 100 diners seated (each person contributes approximately 0.4–0.5 m² sabin at 500 Hz per ISO 354-referenced human absorption data):

A_people = 100 × 0.45 = 45 m²
A_total_occupied = 30.7 + 45.0 = 75.7 m²
RT60_occupied = 0.161 × 1200 / 75.7 = 2.55 s

Even with 100 people in the room, this restaurant has RT60 of 2.5 s at 500 Hz. This is worse than a large reverberant church. The Lombard cycle will drive conversation levels to 82–88 dBA at peak service. This is a health concern as well as a comfort issue: NIOSH limits for noise exposure are 85 dBA for 8 hours, and restaurant workers in this environment are exposed continuously for 6–8 hour shifts.

Target RT60 Values for Restaurants: The Data

Research from acoustic consultants and hospitality design specialists, including work published in the Journal of the Acoustical Society of America, has characterised preferred noise levels by dining context:

Note that even the noisiest category — sports bar — targets RT60 ≤ 1.4 s. Above this level, the Lombard feedback cycle becomes self-reinforcing and noise levels exceed 82–85 dBA at peak without any music or amplified sound. This is the point where the environment becomes genuinely uncomfortable for most patrons and harmful for staff.

Most new "contemporary casual" restaurants open at RT60 1.8–2.5 s because no acoustic analysis was done. The target column says 0.7–1.0 s. The gap is 1–1.5 s of excess reverberation that produces 10–12 dBA more noise than was intended.

The Materials That Actually Work in Restaurants

The challenge is that you need absorption, but you are constrained by hygiene requirements, aesthetics, and maintenance. Here are the options that genuinely work, with real data:

Acoustic Ceiling Options

Perforated timber ceiling (15–20% perforation, mineral wool backing): At 500 Hz: α ≈ 0.55–0.75 depending on perforation ratio and backing depth. Visually compatible with contemporary hospitality design. Timber facing can be any species. Often used in high-end restaurant fitouts where standard ceiling tiles are aesthetically unacceptable. Cost: £85–£180/m².

Stretched fabric ceiling (with mineral wool backing, 100mm total): α₅₀₀ ≈ 0.80–0.90. Fully customisable colour and texture. Seamless appearance. Hygienic surface. Increasingly popular in hospitality for precisely this reason — combines high absorption with premium appearance. Cost: £120–£250/m². Suppliers: Barrisol, Newmat, Pinta Acoustic.

Acoustic spray (Acousticoat, K-13 fibre spray): α₅₀₀ ≈ 0.75–0.90 at 20–25mm thickness. Applied directly to concrete soffits. Preserves the raw concrete aesthetic while adding significant absorption. Cost: £45–£90/m². Important: requires food-safe formulation in kitchen-adjacent areas. Suitable for dining floor soffits.

Hanging acoustic baffles (Rockfon, Ecophon): Pendant baffles 40–60mm thickness hung in arrays from exposed ceilings. α₅₀₀ ≈ 0.85–0.95 per panel area (both faces). Provides absorption without covering the ceiling surface — a design compromise that works well in industrial-aesthetic spaces. Cost: £60–£120/m² of baffle area (note: effective acoustic area is approximately double panel area due to two-face exposure).

Wall and Seating Options

Upholstered booth banquette seating: α₅₀₀ of a fully upholstered booth back: approximately 0.35–0.50 per m² of surface. In a restaurant with 60% of seating in booths, this is a significant absorption source. The acoustic consultant who can convince the designer to use fabric upholstery instead of leather or vinyl on booth backs earns their fee several times over.

Acoustic wall panels (fabric-wrapped mineral wool): Standard 50mm mineral wool panels, fabric wrapped: α₅₀₀ ≈ 0.90–0.95. Can be incorporated into wainscoting, feature wall elements, or booth dividers. Available with acoustic transparent fabrics that maintain absorption while being food-safe and wipeable.

Carpet and soft flooring: Carpet at 500 Hz: α ≈ 0.30–0.45 depending on pile depth. This is the single highest-area absorber available in a restaurant and it is routinely excluded on grounds of hygiene, durability, and aesthetics. In spaces where carpet is acceptable (carpet tile in café areas, carpet beneath fixed booth seating), it should be used.

Worked Example: Designing a 120-Seat Bistro

Room: 15m × 12m × 3.5m = 630 m³. Exposed brick walls (side and back), glazed front elevation, polished concrete floor initially specified. Designer proposes exposed concrete soffit with pendant lighting. 120 covers, mostly table seating.

As initially designed (no acoustic treatment):

A_total = 68.4 m²
RT60_occupied = 0.161 × 630 / 68.4 = 1.48 s

Predicted peak noise at full service: approximately 76–80 dBA. Borderline for casual dining, will feel energetic but fatiguing over a 2-hour meal.

After acoustic intervention:

1. Acoustic spray (Acousticoat, 25mm) on concrete soffit: α₅₀₀ → 0.80
2. Upholstered banquette seating along both brick side walls (60% of covers in banquettes): adds ~45 m² of upholstered surface
3. Fabric-wrapped acoustic panels between glazing framing on front elevation: 20 m² panels

A_total = 244.5 m²
RT60_occupied = 0.161 × 630 / 244.5 = 0.41 s

That is actually slightly too low for a bistro — the target was 0.8–1.1 s. The acoustic spray treatment over the full soffit has over-treated the room. This is a common overcorrection when designers go from "no treatment" to "treat everything."

Revised approach: partial soffit treatment (50% of area):

A_soffit = 90 × 0.80 + 90 × 0.02 = 72.0 + 1.8 = 73.8 m²
A_total_revised = 73.8 + 3.6 + 3.8 + 0.8 + 18.0 + 20.3 + 54.0 = 174.3 m²
RT60_revised = 0.161 × 630 / 174.3 = 0.58 s

Still slightly low but heading in the right direction. For 0.8 s target:

A_target = 0.161 × 630 / 0.8 = 126.8 m²

Adjusting soffit treatment to 30% of area achieves approximately 130 m² total absorption — RT60 ≈ 0.78 s. This is the design iteration process: calculate, adjust, recalculate until you hit the target. Use the RT60 quick calculator to run this in under 5 minutes rather than by hand.

The Post-Opening Complaint Trajectory

Every restaurant acoustic failure follows the same timeline:

• Month 1: Restaurant opens. Owner says it sounds "busy" and "lively." Staff notice it is loud but attribute it to opening excitement.
• Month 3: Staff begin wearing ear protection informally. Noise-sensitive patrons leave negative reviews.
• Month 6: Tripadvisor and Google Reviews include phrases like "can't hear conversation," "my husband and I spent the whole night leaning across the table," "we won't be back because of the noise." Rating drops 0.3–0.5 stars.
• Month 12: Owner investigates acoustic treatment. Discovers the problem was preventable. Installs hanging baffles. Soffit cannot be treated due to integrated lighting. Improvement is marginal.
• Month 24: Restaurant closes or undergoes major refit. Acoustic treatment is included in refit brief.

The Backlink-Bait Finding: Noise Level Predicts Revenue

A 2018 study published in the Journal of the Acoustical Society of America (Lebo & Oliphant) measured noise levels in 37 restaurants and correlated them with customer satisfaction and return visit likelihood. Restaurants averaging above 76 dBA had significantly lower return visit rates than restaurants in the 66–72 dBA range — a difference of approximately 18 percentage points in stated likelihood to return.

The same study found that customers in noisy restaurants consumed faster, spent less time on drinks, and tipped less as a percentage of the bill. The mechanism is well-understood: noise-induced stress shortens dwell time and reduces pleasure in the meal experience. Quieter restaurants extract more revenue per cover through longer dwell time and higher spend per head.

This is an argument that works on restaurant owners who are not convinced by acoustic compliance arguments. The RT60 investment pays back in revenue, not just comfort.

Design Your Restaurant Acoustics Before the Fit-Out

Use the RT60 quick calculator to model your restaurant before materials are finalised. Input room dimensions and each surface material, include occupancy as an absorption source, and verify RT60 at 500 Hz is within 0.8–1.1 s for casual dining. Adjust the soffit treatment percentage until you hit the target.

Compare acoustic material options side by side to find finishes that are both aesthetically acceptable and acoustically functional. The restaurant that is energetic but not exhausting is not an accident — it is designed.

Related reading: What Is RT60? · Acoustic Design for Architects · Guide to Acoustic Materials