Restaurants get louder as they fill because each diner who talks raises the background that everyone else must talk over, and people raise their voices, without meaning to, by roughly half a decibel for every decibel of extra noise. That reflex is called the Lombard effect. Hard surfaces keep each voice in the room longer, so the loop starts from a higher base and climbs faster. There is a limit: a room’s “acoustic capacity”, the most guests it can hold while conversation across a table stays acceptable, which acoustician Jens Holger Rindel estimates as the room volume in cubic meters divided by 20 times its reverberation time in seconds.
This article explains the loop with numbers, shows how to estimate your own room’s capacity, and puts the usual complaint statistics in context with their dates. It is written for owners, chefs, designers and diners who want to know why, not only what to buy. If you want the practical order of fixes, read our companion guide on how to make a loud restaurant quieter.
Why does a restaurant get louder as it fills?
A dining room gets louder as it fills because three things feed each other: more people talking, a background level that rises with every voice, and hard surfaces that keep that sound in the room. Each turn of the loop pushes guests to talk louder, which raises the background again.
Rindel’s 2019 review in Acoustics in Practice, the European Acoustics Association’s journal, describes conversation in a noisy room as “a complicated feed-back situation”. The noise in a restaurant depends on at least four things: the room volume, the reverberation time, the number of people, and the type of gathering. Two of those are fixed when the room is built. The other two change every evening.
The loop has four steps:
- Guests talk. In a typical restaurant, Rindel’s data suggest one person in every three to four is speaking at any moment.
- Hard surfaces keep each voice in the room. Sound reflects off the floor, ceiling, walls and glass, and lingers as reverberation.
- The background rises. All those lingering voices merge into the noise every table must talk over.
- Everyone raises their voice. The Lombard effect lifts each talker’s level, which feeds step 1 at a higher level.
Absorption interrupts step 2, which is why it matters so much. It does not stop anyone talking. It shortens how long each voice lingers, so less of it piles up behind the next one.
The Lombard effect: why everyone talks louder without noticing
The Lombard effect is the involuntary tendency to raise your voice when the noise around you rises. Above a background of about 45 to 60 dBA, depending on the study, talkers add roughly 0.3 to 0.7 dB of voice for each 1 dB of extra noise.
The effect is named after the French ear, nose and throat doctor Étienne Lombard (1869 to 1920), who first reported it. The published numbers vary with the method:
- Rindel’s model starts the effect at a noise level of about 45 dB, with a speech level of 55 dB measured 1 meter in front of the mouth. He cites a review that found slopes from 0.5 to 0.7 dB per dB, and uses 0.5 because it gave the only reasonable fit to measured restaurant and canteen data.
- Bottalico, Piper and Legner (2022) summarize the typical slope as 0.3 to 0.6 dB per dB once noise exceeds 50 dB(A). In their laboratory study of 31 adults over 60, reading aloud over recorded restaurant noise, the slope changed at about 58 dB(A): 0.27 below that point and 0.51 above it. An earlier study by the same lead author with college students found a change point of 57.3 dB(A) and an upper slope of 0.54.
The practical reading is the same in every source. Once the background passes the high 50s, voices climb steeply with it. Below about 45 to 50 dB, the effect is weak or absent, so a room that starts the evening quiet resists the loop for longer: early tables talk at a relaxed level, and the background builds from a lower base.
What the climb does to a conversation
Rindel’s equation for the speaking level is: speech level at 1 meter = 55 + 0.5 × (background noise − 45). The voice rises, but only half as fast as the noise, so the gap between your companion’s voice and the background (the signal-to-noise ratio) shrinks as the room gets louder. Rindel grades conversation quality by that gap.
Raising your voice does not win the gap back, because every other talker raises theirs too. Rindel points out that the Lombard effect leaves the signal-to-noise ratio unchanged on average; what sets it is how many talkers share the room’s absorption. That is why the table below can be read straight from the background level.
| Background noise (dB) | Talker’s level at 1 m (dB) | Signal-to-noise at 1 m (dB) | Rindel’s quality grade |
|---|---|---|---|
| 53 | 59 | +6 | Good |
| 59 | 62 | +3 | Good (lower edge) |
| 65 | 65 | 0 | Satisfactory (lower edge) |
| 71 | 68 | −3 | Sufficient (lower edge) |
| 77 | 71 | −6 | Insufficient |
| 83 | 74 | −9 | Insufficient (lower edge) |
In the vocal effort scale Rindel adapts from earlier research, 60 dB at 1 meter is “normal, raised”, 66 dB is “raised” and 72 dB is “loud”. At a 71 dB background, every talker in the room is already between raised and loud, and the conversation is only just sufficient. That is why a busy room is tiring: guests work harder and still hear less.
How reverberation keeps the noise in the room
Reverberation is sound that keeps reflecting after it leaves the talker. In a hard room it adds up into a steady background, and that background, not the voice of the person beside you, is what you must talk over.
Acoustical consultant Marshall Long explained the split in a 2005 feature for Acoustics Today, the Acoustical Society of America’s magazine. The voice of the person you are listening to arrives as direct sound, which falls by 6 dB each time the distance doubles. The rest of the room arrives as reverberant sound, which is roughly the same everywhere and depends on the total sound power of all the talkers and on the room’s absorption.
Long’s illustration is a party in a living room with a carpeted floor, hard walls and ceiling. With only the two hosts talking, they could understand each other at up to 3.9 meters (13 ft). When a second conversation started, the reverberant level rose 3 dB and that distance fell to 2.7 meters (9 ft). Two more couples brought it to 1.9 meters (6 ft); four more, to 1.3 meters (4 ft). “In practice,” he wrote, people “may choose not just to move closer, but also to talk louder,” which raises the background and forces everyone to raise their voices, so that “at the end of the evening they all go home with sore throats.”
Then he applied it to a restaurant. In a hard 45 by 45 ft room with little absorption, one talker at each of 20 tables pushed the reverberant noise to 76 dB, too loud for an intelligible conversation across the table. Covering the ceiling with a material that absorbs 90 percent of the sound reaching it brought the level to 66 dBA, which Long called just low enough for a cross-table conversation. His conclusion: more absorption raises the signal-to-noise ratio, so more people can talk comfortably “before the increasing-volume spiral begins to kick in.”
Why noise climbs about 6 dB each time the crowd doubles
Without the Lombard effect, doubling the number of talkers would add 3 dB, because two equal sound sources add up to 3 dB more than one. With the Lombard effect, each extra 3 dB makes everyone talk louder, and the loop settles about 6 dB higher. Rindel’s model and older measurements both show the 6 dB step.
Rindel’s simple prediction equation, validated against food courts, a canteen, a reception and a banquet, gives the background level from speech as: L = 93 − 20 × log10(A × g ÷ N), where A is the room’s absorption in square meters, N is the number of people and g is the group size (people per active talker, typically 3.5 in restaurants). The 20 in front of the logarithm is where the Lombard effect shows up. It means:
- Doubling the guests at the same absorption adds about 6 dB. Rindel notes that an earlier study of dining rooms by Gardner found the same.
- Doubling the absorption at the same occupancy removes about 6 dB, twice what simple room acoustics would predict.
A quick way to see where the 6 dB comes from: if the Lombard slope is 0.5, half of every rise feeds back into voices, so the first 3 dB from doubling the crowd grows to 3 ÷ (1 − 0.5) = 6 dB once the loop settles.
The model is honest about its limits. The group size is the main uncertainty: if the real group size ranges from 2.5 to 5, predictions move by 6 dB in total, so Rindel gives the method an uncertainty of about ±3 dB. He also warns that it may not apply to small rooms holding fewer than about 30 people.
Rooms remember
Two of Rindel’s case studies show that the loop has momentum. In a canteen measured at lunchtime, the noise level near the end, with about 50 people left, was about 5 dB higher than with the same number of people at the start, because more of them were talking. At a university banquet, the noise rose 15 to 20 dB in the first half hour and then stayed fairly stable for hours. The late-evening level in your room is not just the sum of the people in it; it is also the mood they built.
Why the climb levels off
The loop does not run forever. Three things slow it down.
- Voices have a ceiling. Rindel puts shouting at 84 to 90 dB at 1 meter, and notes that speech at very high vocal effort, above about 75 dB, can be harder to understand than speech at lower effort. Past a point, talking louder stops helping.
- People absorb sound. Clothed bodies add absorption, about 0.2 to 0.5 square meters per person depending on clothing in Rindel’s model. He says that contribution should be counted once the noise passes about 73 dB, which is one reason very full rooms do not keep climbing at the same rate.
- Guests lean in. Cutting the distance from 1 meter to 0.7 meters gains 3 dB of signal-to-noise, and moving to 0.5 meters gains another 3 dB. Rindel calls that the obvious way to keep talking in a room that is too loud, but adds that it does not change the noise level, “which makes the environment unpleasant for a longer stay.”
What is acoustic capacity?
Acoustic capacity is the maximum number of people a room can hold while conversation at 1 meter (about a four-top’s width) stays at least “sufficient”. Rindel defines it in his 2019 review and estimates it as N = V ÷ (20 × T), with V the volume in cubic meters and T the furnished, unoccupied reverberation time at mid frequencies (500 to 1,000 Hz).
The threshold behind it is the 71 dB background from the table above, where the signal-to-noise ratio at 1 meter falls to −3 dB. The formula assumes a group size of 3.5 and absorption of 0.35 square meters per person, the middle of the 0.2 to 0.5 range he gives for clothed people.
In feet: one cubic meter is about 35.3 cubic feet, so the same rule is N ≈ V ÷ (706 × T), with V in cubic feet.
Capacity follows absorption
Put Sabine’s formula into Rindel’s and the volume cancels out: the capacity comes to about the room’s total absorption in square meters divided by 3.2. In US units, that is roughly total sabins ÷ 34. The fictional example room below, with 662.8 sabins, comes out at 662.8 ÷ 34.4 = 19 guests by this route too.
Volume still matters, because a taller room holds more absorption at the same reverberation time. Suppose two dining rooms share a 2,000 sq ft floor and a 1.0 second reverberation time, one with a 9 ft ceiling and one with a 16 ft ceiling. The first has 18,000 cu ft and needs 0.049 × 18,000 ÷ 1.0 = 882 sabins to reach 1.0 second, for a capacity of about 18,000 ÷ 706 = 25 guests. The second has 32,000 cu ft, needs 1,568 sabins, and holds about 45 guests. The taller room also has more upper wall and more hanging depth to carry that absorption out of reach. That is why Rindel lists a big volume first among his design recommendations.
Capacity depends on how far apart people sit
A 1 meter conversation is a standard, not a law. Rindel’s distance table shows how much louder a room may get before conversation falls below “sufficient” at other distances, and he maps them to common table shapes.
| Conversation distance | Typical seat | Highest background for “sufficient” |
|---|---|---|
| 0.5 m (1.6 ft) | The person next to you | 83 dB |
| 0.7 m (2.3 ft) | Across a small table for two | 77 dB |
| 1.0 m (3.3 ft) | Across a square table for four | 71 dB |
| 1.4 m (4.6 ft) | Across a round table for six | 65 dB |
| 2.0 m (6.6 ft) | Across a round banquet table for ten | 59 dB |
This is why big round tables are hard work in a busy room. Rindel notes that a conversation across a table for ten would need a background of no more than 59 dB, and at 77 dB guests can only talk to the person beside them.
Full restaurants usually run over capacity
Rindel’s own assessment is blunt: “When a restaurant is fully occupied, it is typical that the acoustic capacity is exceeded by a factor of 2 or more.” His suggested remedies follow from the formula: build as much volume as possible (a high ceiling helps), apply absorption to the ceiling and often part of the walls, and do not set the seating plan more than 25 percent above the acoustic capacity.
He also turns the formula into a design requirement. If you plan for 80 percent occupancy, the reverberation time should be no more than about 0.063 × V ÷ N seconds, with V in cubic meters and N the number of seats. In feet: about V ÷ (560 × N), with V in cubic feet.
How many guests can the example dining room take?
Applied to the site’s fictional 60-seat example dining room, Rindel’s formula gives an acoustic capacity of about 8 guests as built and about 19 with the planned panels. The panels more than double the number of people the room can hold before four-top conversation suffers, but the room still runs well over capacity when it is full.
Example (fictional). Not a Fibson project. The site’s example dining room: 60 seats, 40 by 30 ft, 11 ft ceiling, volume 13,200 cu ft (373.8 cubic meters). Its Sabine estimate (RT60 = 0.049 × V ÷ A, with planning-assumption coefficients, not test results) is 2.34 seconds as built and 0.98 seconds with 16 ceiling clouds, 14 fabric wall panels and 4 printed art panels.
Acoustic capacity, N = V ÷ (20 × T):
- As built: 373.8 ÷ (20 × 2.34) = 373.8 ÷ 46.8 = 8.0 guests.
- Clouds only (1.28 s): 373.8 ÷ 25.6 = 14.6, about 15 guests.
- All panel groups (0.98 s): 373.8 ÷ 19.6 = 19.1, about 19 guests.
In feet, as a check: 13,200 ÷ (706 × 0.98) = 13,200 ÷ 691.9 = 19.1 guests.
Predicted crowd noise at 60 guests, from Rindel’s equations with g = 3.5 and 0.35 square meters of absorption per person:
- With panels: A = 0.16 × 373.8 ÷ 0.98 + 0.35 × 60 = 61.0 + 21.0 = 82.0 square meters. Then 82.0 × 3.5 ÷ 60 = 4.78, and 20 × log10(4.78) = 13.6. L = 93 − 13.6 = 79.4 dB.
- As built: A = 25.6 + 21.0 = 46.6 square meters; 46.6 × 3.5 ÷ 60 = 2.72; 20 × log10(2.72) = 8.7; L = 84.3 dB.
Guests the room holds before conversation falls below “sufficient”, by seat type:
| Seat type | Limit | As built | With all panels |
|---|---|---|---|
| Four-top, 1.0 m | 71 dB | about 8 | about 19 |
| Two-top, 0.7 m | 77 dB | about 18 | about 42 |
| Next to each other, 0.5 m | 83 dB | about 46 | more than 60 |
What 60 seats would need at 80 percent occupancy: T ≤ 0.063 × 373.8 ÷ 60 = 0.39 seconds, which needs about 646.8 ÷ 0.39 = 1,660 sabins of absorption, roughly two and a half times the 662.8 sabins in the panel plan.
All predictions carry Rindel’s stated uncertainty of about ±3 dB.
How to read the example
Three lessons carry over to real rooms.
- Absorption buys guests, not silence. Acoustic capacity is proportional to absorption, because a shorter reverberation time means more absorption. More than doubling the absorption more than doubled the capacity. It did not make the full room quiet; it moved the point where conversation breaks down from a handful of tables to a third of the seats.
- Small tables survive longer. With panels, two-tops stay workable up to about 42 guests in this model, while four-tops struggle from about 19. A mix of small tables helps a room that runs near full.
- Sixty talkers need a lot of absorption. Rindel’s requirement for 60 seats at 80 percent occupancy works out at about 1,660 sabins in this room, roughly two and a half times the panel plan. The ceiling and walls together offer about 2,520 sq ft of surface, so meeting it with panels alone would mean covering most of the room. That is why music, table size and seat count matter alongside absorption.
The site’s 1 second target for this room is a planning choice that balances cost, looks and benefit; no code sets one for restaurants. You can switch each panel group on and off in the example room walkthrough, and run your own room through the reverberation time calculator. Our article on reverberation time targets explains where published targets come from.
Is it the decor? What changed in restaurant design
Decor is part of the answer, but not all of it. Hard, open interiors make the loop start earlier, loud music adds to it, and some venues want a lively room on purpose.
Hard rooms are a design choice
Long put the design side bluntly in 2005: “Form follows fashion rather than function in the world of architecture.” In hard-surfaced restaurants, he wrote, it is very difficult to hold a normal conversation across a table, “yet for some unfathomable reason countless restaurants are designed in this way.” He also described the incentive: a hard-edged room photographs well in an architecture magazine, even when it is noisy. His closing line is often quoted by acousticians: “Ultimately it is easier to design new materials that solve the architects’ problems than it is to design new architects.”
Gregory Scott’s 2017 paper for the Acoustical Society of America lists the usual suspects: background music, and design elements such as open kitchens, stripped-down or hard surfaces, and fewer tablecloths, less carpeting and less paneling to absorb sound. Bottalico and colleagues add a cultural note: before the 1990s, by their account of the literature, “excessive” restaurant noise was not viewed negatively by diners and critics because it was supposed to add to the atmosphere.
Music raises the floor
Music is part of the background, so the Lombard effect responds to it too. Rindel recommends that background music stay at about 60 to 65 dB at most, and suggests 60 dB if you want conversation to stay “satisfactory”. With music at 65 dB, his model gives “sufficient” conversation only when the room holds fewer than about 70 percent of its acoustic capacity. Foreground music, meant to be listened to, runs 75 to 90 dB, and guests are not expected to talk over it.
Some rooms are loud on purpose
Not every owner wants a quiet room. Rindel’s review summarizes a study by Guéguen and colleagues in which bar customers drank more with louder music: an average of 3.7 drinks at 88 to 91 dB against 2.6 at the usual 72 to 75 dB, across 60 people. Another study he cites found food tasted less intense in loud noise. His conclusion is that a fine restaurant should keep noise low to protect taste, while a bar may accept, or even want, a higher level. Our guide to bar and taproom acoustics covers rooms where energy is part of the product.
Is it the alcohol?
Many owners assume rooms get louder as the drinks flow. Rindel’s review found no proof in the scientific literature that the noise of a party rises with the amount of alcohol consumed. The evidence he cites runs the other way: louder music led bar customers to drink more and faster. His model does treat the type of gathering as a variable, since how lively a group is, how well people know each other, their age and their drinking all change how many of them talk at once. Alcohol may change the group, but the room decides how far each voice carries.
What diners report, and when
Noise has topped diner complaint surveys for years, but the most quoted numbers are from 2016 to 2018. Measurement studies tell a consistent story about levels, and each has limits worth knowing.
| Year | Source | What was studied | Finding | Keep in mind |
|---|---|---|---|---|
| 2007 | Hodgson and colleagues, J Acoust Soc Am | Ten eating establishments | Diners and staff exposed to 45 to 82 dB(A) | Small sample |
| 2014 | To and Chung, Noise and Health | Peak-hour levels in Hong Kong restaurants | 66.7 to 82.6 dBA; levels depended on occupancy density only | One city; peak hours only |
| 2016 | Consumer Reports | Survey respondents’ visits to chain restaurants | Noise (loud customers, blaring music) was the biggest complaint: 16 percent of dining experiences on average, against 12 percent for poor service | Chains only |
| 2017 | Scott, Acoustical Society of America | Crowd-sourced readings at more than 1,800 Manhattan venues, 2015 to 2017 | Restaurants averaged 78 dBA, bars 81 dBA | Phone app data, prime time only |
| 2017 | Spira-Cohen and colleagues, NYC health department | 59 loud venues on Thursday to Saturday evenings | Restaurants and bars averaged 91 dBA | Venues chosen for being loud |
| 2018 | Zagat Dining Trends survey | Nearly 13,000 avid diners | Noise most bothersome at 24 percent, just ahead of service at 23 percent | Self-selected respondents |
The most quoted survey numbers are now eight to ten years old, and they come from different questions asked of different groups. Treat them as a consistent snapshot, not a trend line: noise was at or near the top of diners’ complaints in every survey above, and the measurements explain why.
The Hong Kong result deserves the most attention from owners. Across restaurant types, the measured noise depended on how densely the room was occupied and on nothing else the researchers recorded, and their model showed people raising their voices as density increased. That is the loop in the wild.
Who loses out first?
Older diners and people with hearing loss are the first to give up on a loud room, and they reach that point at levels many younger guests barely notice.
What the research shows
- Older adults. In Bottalico’s study, the older participants’ reported disturbance rose steeply above 51.6 dB(A). Their willingness to spend money fell fastest between 54.6 and 62.0 dB(A), at about 4 percent of the budget per decibel, and above 62 dB(A) they said they would spend less than a quarter of it. Participants with hearing loss were more disturbed, and less willing to spend time and money, than those with normal hearing at the same noise level.
- People with hearing loss. In a Norwegian survey cited by Rindel, 51 percent of 251 hearing-impaired respondents said they often or always had difficulty holding a conversation in canteens, restaurants and cafes because of noise from speech. Including “sometimes”, the figure rose to 88 percent.
- Non-native listeners. Rindel cites the international standard ISO 9921, which says people with a slight hearing disorder (in general the elderly) and non-native listeners need a signal-to-noise ratio about 3 dB higher.
What 3 dB means for capacity
A 3 dB higher signal-to-noise requirement looks small, but it compounds. In Rindel’s equations, the signal-to-noise ratio at a given distance depends on the absorption available per talker, and each extra 3 dB needs twice as much of it. In the same room, that means half as many guests, which also shows up as a background about 6 dB lower. For a hearing-impaired or non-native diner, the room’s acoustic capacity is roughly halved. In the fictional example room with panels, that is about 9 guests at a four-top distance instead of 19.
Rindel frames this as universal design: the idea that public buildings should be accessible to everyone, including their acoustic conditions, not only their doors and ramps. A room that works for a guest with mild hearing loss works better for everyone else at the table too.
Staff hear the room longest of all. For workplace limits and what they mean for servers and bartenders, see the staff section of our guide to reducing restaurant noise.
How to watch the loop in your own room
You can see the loop in your own dining room with a phone and one evening of notes. Log the sound level against the number of guests, and look at what happens each time the room doubles.
- Pick one spot in the middle of the seating, away from speakers, the bar and the kitchen pass. Use a sound level meter app set to A-weighting (dBA).
- Fix the music at one setting for the whole evening, so the crowd is the only thing changing.
- Every 15 to 30 minutes, take a reading and write down the number of guests seated, from the host stand.
- Compare doublings. Once the room is past its first few tables, each doubling of guests should add something near 6 dB if the loop is running. Much more suggests music or a source changed; much less suggests a well-absorbing room or quiet groups.
- Check the late evening. Take a reading when the room has emptied back to an earlier guest count. If it is louder than before at the same count, the room has momentum.
Example (fictional). These readings are invented to show the method, not measurements of a real room. A 64-seat dining room with a plaster ceiling and a tiled floor logs one Saturday with the music fixed at one setting.
| Time | Guests seated | Reading |
|---|---|---|
| 6:00 pm | 12 | 68 dBA |
| 6:45 pm | 24 | 74 dBA |
| 7:45 pm | 48 | 80 dBA |
| 9:30 pm | 24 | 77 dBA |
Each doubling from 12 to 24 to 48 guests added 6 dB, the signature of the Lombard loop in a hard room. At 9:30 pm, with the same 24 guests as at 6:45 pm, the room is 3 dB louder, the kind of momentum Rindel’s canteen showed. The room passes 71 dBA somewhere between 12 and 24 guests, so its practical capacity for four-top conversation is under a third of its 64 seats.
Readings from a phone are approximate, but the pattern matters more than the absolute number. Two or three logged evenings tell you whether your problem is the crowd and the room, or something you can switch off.
What the loop means for owners: the levers, ranked
Every lever in Rindel’s model is a variable you can name: volume, absorption, guests, group behavior, music and table size. After the room is built, absorption and music are the cheapest to change; seats and tables are business decisions.
- Absorption, mostly on the ceiling. It is the only lever that raises capacity without losing covers. Each doubling of absorption roughly doubles capacity and, in an occupied room, removes about 6 dB of crowd noise. The ceiling over the seating is usually the largest free surface.
- Music level. Keep background music at or below 60 to 65 dB. Above that, it raises voices and cuts the number of guests who can talk comfortably.
- Table mix. Smaller tables keep conversation distances short. In a room that runs near capacity, two-tops and four-tops cope better than rounds for six or ten.
- Seat count. Rindel suggests seating no more than 25 percent above acoustic capacity. Few owners will cut seats, but knowing your room’s capacity tells you which nights will be hard and why.
- Volume. In a new build or a gut renovation, a higher ceiling lets the room carry more absorption at the same reverberation time, out of reach of guests. Treat ceiling height as an acoustic decision, not only an architectural one.
How seating layout, booths and zoning change the picture at the same number of covers is covered in our sibling article on table layout and restaurant noise.
Estimate your own room in four steps
- Measure the volume. Length × width × average ceiling height, in cubic feet.
- Estimate the reverberation time of the furnished, empty room with the reverberation time calculator, from your surface finishes.
- Divide. Acoustic capacity ≈ volume in cubic feet ÷ (706 × reverberation time in seconds).
- Compare with your seats. If your seat count is two or three times the result, the loop will run on busy nights. Rindel suggests aiming for no more than 1.25 times.
If you measure your room’s sound levels at different occupancies, our decibel addition calculator helps you separate the music from the crowd. For what a dining room panel plan usually includes, see the restaurants, cafes and bars page. Fibson designs and makes acoustic panels to order and plans each room before anything is ordered; if you want your capacity estimate checked against a plan, send us your room with its dimensions, finishes and seat count.
