Think of background noise as the floor of a room's acoustic environment. Everything you try to hear — speech, music, a presentation — sits on top of that floor. If the floor is too high, the things on top get crowded out. If it tilts, with one end higher than the other, certain sounds become muddy while others turn shrill. Noise criteria curves are the building codes for that floor. They tell you exactly how high it is allowed to be at each frequency, the way speed limits tell you how fast you can drive on each type of road. A residential street gets 30 km/h. A motorway gets 120 km/h. A concert hall gets NC-15. A restaurant gets NC-40. Same idea, different domain.
The complication is that three separate systems exist for setting those limits — NC, NR, and RC — and they do not agree on which frequencies matter, how to rate the result, or even what country you should be using them in. If you are an architect specifying HVAC noise targets, a mechanical engineer designing ductwork, or a consultant writing a performance specification, choosing the wrong system (or confusing the numbers between systems) can mean a space that either costs too much to build or fails commissioning.
This guide explains how each system works, where each one applies, and how to pick the right one for your project.
Why Background Noise Matters More Than You Think
Background noise in occupied buildings comes primarily from three sources: HVAC systems (ductwork, diffusers, air handling units), external noise intrusion (traffic, aircraft, adjacent spaces), and building services (lifts, plumbing, electrical transformers). Of these, HVAC is dominant in most modern buildings. A well-sealed facade can attenuate external noise by 35-45 dB, but the air conditioning system sits inside the building envelope, delivering its noise directly into every served space through the ductwork that was designed to deliver conditioned air.
The consequences of excessive background noise are well documented and consistently underestimated.
Speech Intelligibility Degradation
The Speech Transmission Index (STI), defined in IEC 60268-16:2020, quantifies how much of a spoken message reaches the listener intact. STI ranges from 0.00 (completely unintelligible) to 1.00 (perfect transmission). Background noise directly reduces STI by masking the modulation patterns that carry speech information. A private office with NC-25 background noise might achieve STI of 0.75 (good). Raise the background noise to NC-40 — still within acceptable limits for an open office — and STI at the same distance drops to 0.55 (fair). The speaker has not changed. The room has not changed. The noise floor ate the difference.
Productivity and Cognitive Load
Banbury and Berry (2005) measured a 15-28% reduction in cognitive task performance when subjects were exposed to background noise containing intelligible speech. Critically, the effect persists even when the noise is below conscious awareness thresholds. The human auditory system processes speech automatically. You cannot choose not to hear a conversation at your neighbour's desk any more than you can choose not to see a flashing light in your peripheral vision. The processing happens before conscious attention gets a vote.
Annoyance and Spectral Character
Not all background noise is equally annoying. A broadband, neutral-spectrum noise at 40 dBA is far less disturbing than a tonal noise at 35 dBA with a prominent low-frequency hum or a high-frequency whine. This is precisely why the RC system was developed — NC and NR tell you the level, but they say nothing about the spectral shape. A room can pass NC-35 while containing a 63 Hz rumble from a rooftop air handling unit that makes every occupant uneasy without being able to articulate why.
NC (Noise Criteria) — The American System
History and Development
The Noise Criteria system was developed by Leo Beranek in 1957 at Bolt Beranek and Newman (BBN), the consulting firm that would later help design the acoustics for the United Nations General Assembly Hall and Lincoln Center. Beranek needed a practical method for specifying acceptable background noise levels in buildings — something that could bridge the gap between acoustic consultants who thought in octave-band sound pressure levels and architects who needed a single number to put in a specification.
The result was a family of curves, each identified by a single number (NC-15, NC-20, NC-25, and so on in increments of 5), plotted across octave-band center frequencies from 63 Hz to 8000 Hz. The curves are not flat. They slope downward from low frequencies to high frequencies, reflecting the ear's greater sensitivity to high-frequency sound and greater tolerance of low-frequency rumble. An NC-35 curve permits approximately 60 dB SPL at 63 Hz but only 36 dB SPL at 8000 Hz.
How NC Rating Works
To determine the NC rating of a room, you measure the sound pressure level in each octave band from 63 Hz through 8000 Hz (eight bands total) and plot the results against the family of NC curves. The NC rating is the lowest NC curve that is not exceeded by any measured octave band. This is sometimes called the "tangent method" — the NC value is determined by the single octave band that just touches the highest NC curve.
For example, if a room measures 52 dB at 125 Hz, 45 dB at 250 Hz, 38 dB at 500 Hz, 33 dB at 1000 Hz, 29 dB at 2000 Hz, and 27 dB at 4000 Hz, and every one of those values falls below the NC-35 curve except the 125 Hz reading which touches the NC-40 curve, the room rates NC-40. One octave band sets the rating. The rest are irrelevant to the final number.
The NC Limitation
This tangent method is both the strength and the weakness of NC. The strength is simplicity: a single number that anyone can understand and specify. The weakness is that NC tells you nothing about the spectral shape of the noise. Two rooms can both rate NC-35 while sounding completely different. One might have a flat, neutral spectrum that occupants find pleasant. The other might have all its energy concentrated in a single low-frequency band, producing a rumble that causes headaches. NC-35 is NC-35 — the number does not distinguish between them.
The Standard
NC is defined in ANSI/ASA S12.2-2019, "Criteria for Evaluating Room Noise." This standard also defines the RC system (discussed below), and provides extensive tables of recommended NC values by room type.
Common NC Targets
| Room Type | Recommended NC | Notes |
|---|---|---|
| Concert hall | 15-20 | Background noise must not mask pianissimo passages |
| Recording studio | 15-20 | Microphones amplify everything; floor noise must be inaudible |
| Theatre / auditorium | 20-25 | Unamplified speech requires very low background levels |
| Courtroom | 25-30 | Speech intelligibility is a legal requirement |
| Private office | 30-35 | Conversations should be confidential at normal voice levels |
| Conference room | 25-30 | Teleconference systems are sensitive to background noise |
| Classroom | 25-30 | ANSI S12.60 mandates 35 dBA (approximately NC-28) |
| Open plan office | 35-40 | Some masking is beneficial for speech privacy |
| Hospital ward | 30-35 | Patient recovery is adversely affected by noise |
| Lobby / corridor | 40-45 | Transient spaces tolerate higher levels |
| Restaurant / cafeteria | 40-45 | Social noise makes high background levels acceptable |
NR (Noise Rating) — The European and ISO System
History and Development
The Noise Rating system was developed by the International Organization for Standardization and is documented in ISO 1996. Where NC was born in American consulting practice, NR was born in the international standards framework and became the default specification system across Europe, the UK, much of Asia, and Australia. If your project is in London, Berlin, Singapore, or Sydney, the mechanical engineer's specification will almost certainly use NR, not NC.
NR curves look similar to NC curves — they slope downward from low frequencies to high, and they are identified by a single number — but there are important differences.
Frequency Range
NR covers eight octave bands from 31.5 Hz to 8000 Hz. NC starts at 63 Hz. That extra low-frequency octave band at 31.5 Hz is significant because many HVAC noise problems manifest below 63 Hz. Large centrifugal fans, cooling towers, and rooftop packaged units can generate substantial energy at 31.5 Hz that the NC system simply does not see. A room that passes NC-30 might fail NR-30 because of a 31.5 Hz component that NC never measured.
How NR Rating Works
The method is the same tangent approach as NC. Measure the octave-band spectrum, plot it against the NR curves, and the NR rating is set by the highest curve that any single octave band touches. The numerical values of the NR curves differ slightly from NC curves at corresponding frequencies, so NR-35 and NC-35 are not equivalent specifications. In general, NR values run approximately 2-5 points higher than the corresponding NC value for the same measured spectrum, depending on the spectral shape. This means that specifying "NR-35" is roughly comparable to specifying "NC-30 to NC-33" — but the correspondence is not exact enough to substitute one for the other without checking.
Where NR Is Used
NR is the standard specification system in the following jurisdictions and frameworks:
- United Kingdom: BB93 (school acoustics), Building Regulations Approved Document E, BREEAM assessments
- European Union: Most national building codes reference NR through ISO 1996
- Australia: NCC 2022 references NR values alongside AS 2107 recommendations
- Singapore: Code of Practice for Environmental Control (noise)
- Middle East: Most specifications follow British or ISO practice
Common NR Targets
| Room Type | Recommended NR | Notes |
|---|---|---|
| Concert hall | 20-25 | Slightly higher numerically than NC equivalent |
| Recording studio | 15-20 | Critical listening demands the lowest achievable level |
| Theatre / auditorium | 25-30 | Includes assessment at 31.5 Hz |
| Courtroom | 25-30 | Legal proceedings require clear speech |
| Private office | 30-35 | Confidential speech at normal voice levels |
| Conference room | 25-30 | Teleconferencing and video calls |
| Classroom | 25-30 | BB93 specifies maximum ambient noise levels in NR |
| Open plan office | 35-40 | Background masking reduces distraction radius |
| Hospital ward | 30-35 | WHO guidelines for patient environments |
| Lobby / corridor | 40-45 | Short-stay spaces with no speech privacy requirement |
| Restaurant / cafeteria | 40-45 | Social venues tolerate higher levels |
RC (Room Criteria) — The Refined American System
History and Development
By the 1990s, acoustic consultants in North America were growing increasingly frustrated with NC's inability to describe the character of background noise. A room could meet its NC specification perfectly and still draw complaints because the noise had an unpleasant spectral quality — a low-frequency rumble from a cooling tower, a mid-frequency drone from a transformer, or a high-frequency hiss from diffusers running at excessive face velocity.
Warren Blazier, working with ASHRAE's Technical Committee 2.6 (Sound and Vibration), developed the Room Criteria (RC) system in 1997 to address this gap. RC was adopted into ANSI/ASA S12.2 alongside NC and is now the preferred criterion in ASHRAE applications.
How RC Works
RC takes a fundamentally different approach from NC and NR. Instead of just producing a level number, it produces a level number and a spectral quality descriptor.
Step 1: Calculate the RC level. Average the measured sound pressure levels in the 500 Hz, 1000 Hz, and 2000 Hz octave bands. This three-band average is the RC level. Unlike NC and NR, the RC level is not determined by a tangent to a family of curves. It is a simple arithmetic mean of three mid-frequency octave bands.
Step 2: Determine the spectral quality. Compare the measured spectrum against the RC reference spectrum (which is the straight line passing through the RC level at 1000 Hz with a slope of -5 dB per octave). If the low-frequency octave bands (16 Hz to 250 Hz) exceed the reference by more than 5 dB, the noise has excessive low-frequency content and receives the qualifier (R) for "rumble." If the high-frequency bands (2000 Hz to 4000 Hz) exceed the reference by more than 3 dB, it receives (H) for "hiss." If neither threshold is exceeded, the spectrum is (N) for "neutral."
The result is expressed as, for example, RC-33(N) — meaning the room has a background noise level corresponding to RC-33 with a neutral, balanced spectrum. Or RC-38(R) — meaning the level is 38 with excessive low-frequency rumble. Or RC-30(RH) — meaning the level is 30 but the spectrum has both rumble and hiss problems.
The Frequency Range Advantage
RC assesses frequencies from 16 Hz to 4000 Hz. This is a much wider range than NC (63 Hz to 8000 Hz) at the low end and a narrower range at the high end. The extension down to 16 Hz is critical because the most common source of occupant complaint in modern buildings is low-frequency noise from HVAC equipment — noise that NC cannot evaluate because its lowest band starts at 63 Hz.
A rooftop air handling unit producing significant energy at 31.5 Hz will show up in an RC assessment as an (R) rumble qualifier. In an NC assessment, that same energy is invisible. The room passes NC and the occupants complain. RC catches it.
Why RC Is Not Universal
If RC is objectively better at characterizing noise quality, why has it not replaced NC and NR everywhere? Three reasons.
First, inertia. NC has been in specifications for almost 70 years. NR has been in ISO standards for decades. Architects, engineers, and code officials know these systems. Changing specifications means changing institutional knowledge, contract templates, and compliance procedures. That takes decades even when the technical case is clear.
Second, measurement practicality. RC requires measurement down to 16 Hz, which demands instrumentation capable of accurate readings at very low frequencies. Many sound level meters used in field commissioning are not reliable below 31.5 Hz, making RC assessments more expensive and technically demanding than NC or NR assessments.
Third, international recognition. NR is embedded in ISO standards and national building codes across dozens of countries. RC is an American standard (ANSI S12.2) with limited international adoption. A consultant working on a project in the UK cannot specify RC — the building regulations do not recognise it.
Head-to-Head Comparison
| Feature | NC | NR | RC |
|---|---|---|---|
| Origin | USA (Beranek, 1957) | ISO (Europe) | USA (Blazier, 1997) |
| Standard | ANSI/ASA S12.2 | ISO 1996 | ANSI/ASA S12.2 |
| Frequency range | 63-8000 Hz | 31.5-8000 Hz | 16-4000 Hz |
| Number of octave bands | 8 | 8 | 8 |
| Rating method | Tangent to curve family | Tangent to curve family | Average of 500/1000/2000 Hz |
| Spectral quality indicator | No | No | Yes (N, R, H, or combinations) |
| Detects low-frequency rumble | Poorly (starts at 63 Hz) | Partially (starts at 31.5 Hz) | Yes (starts at 16 Hz) |
| Primary regions of use | North America | Europe, UK, Asia, Australia | North America (ASHRAE) |
| Ease of field measurement | High | High | Moderate (requires 16 Hz capability) |
| Recommended by ASHRAE | Supported | Not referenced | Preferred |
Recommended Values by Room Type
The following table consolidates the major standards and guidelines into a single reference. Values are expressed as ranges; the lower end represents the ideal target, and the upper end represents the maximum acceptable level. Projects targeting premium acoustic quality (WELL v2 Platinum, high-end commercial, critical listening) should aim for the lower end.
| Room Type | NC | NR | RC |
|---|---|---|---|
| Concert hall / recital room | 15-20 | 20-25 | 15-20(N) |
| Recording studio / broadcast | 15-20 | 15-20 | 15-20(N) |
| Theatre / auditorium (unamplified) | 20-25 | 25-30 | 20-25(N) |
| Courtroom | 25-30 | 25-30 | 25-30(N) |
| Classroom (ANSI S12.60 / BB93) | 25-30 | 25-30 | 25-30(N) |
| Private office | 30-35 | 30-35 | 30-35(N) |
| Conference / meeting room | 25-30 | 25-30 | 25-30(N) |
| Teleconference / video room | 20-25 | 25-30 | 25(N) |
| Open plan office | 35-40 | 35-40 | 35-40(N) |
| Hospital ward / patient room | 30-35 | 30-35 | 30-35(N) |
| Library / reading room | 30-35 | 30-35 | 30-35(N) |
| Hotel guest room | 30-35 | 30-35 | 30-35(N) |
| Lobby / reception | 40-45 | 40-45 | 40-45(N) |
| Restaurant / cafeteria | 40-45 | 40-45 | 40-45(N) |
| Retail / shopping mall | 40-45 | 40-45 | 40-45(N) |
| Gymnasium / sports facility | 40-50 | 40-50 | 40-50(N) |
Note that the (N) qualifier on all RC values in this table is intentional. No room should be designed to have a rumble or hiss character. If your RC assessment returns an (R) or (H) qualifier at any level, the HVAC system needs remediation regardless of whether the numerical level is acceptable.
Which System Should You Use?
The answer depends almost entirely on project location and the applicable standards framework.
Use NC if: The project is in the United States or Canada, the specification references ANSI S12.2, or the client's acoustic consultant works in American practice. NC is understood by every HVAC contractor and commissioning agent in North America.
Use NR if: The project is in the United Kingdom, European Union, Australia, Singapore, or any jurisdiction where the building code or planning conditions reference ISO 1996 or NR directly. Using NC in a London office fit-out will confuse the mechanical engineer and may not satisfy the building control officer.
Use RC if: The project is in North America and the mechanical system is complex enough that spectral quality matters — large open plan offices, broadcast facilities, performing arts centres, or any space where ASHRAE guidelines apply. RC is particularly valuable when specifying noise limits for large central plant serving multiple zones, because the (R)/(H)/(N) qualifier gives the HVAC engineer actionable information about which end of the spectrum to control.
Use more than one if: High-performance projects often specify both a level criterion (NC or NR) and a quality criterion (RC). For example, "NC-30 with no RC rumble qualifier" gives the HVAC engineer a familiar target number while protecting against the spectral quality gap that NC alone cannot address.
How to Reduce Background Noise to Meet Your Target
Knowing the target is the first step. Achieving it requires coordinated effort between the acoustic consultant, the mechanical engineer, and the architect. The following strategies address the most common sources of failure.
HVAC Duct Design
Ductwork is a highway for noise. Sound generated by the air handling unit propagates through the duct system and radiates into rooms through diffusers and grilles. The three primary control strategies are:
- Duct lining: Internal acoustic lining (typically 25 mm or 50 mm mineral wool) absorbs sound energy as it travels through the duct. Effective at mid and high frequencies (500 Hz and above). Less effective below 250 Hz, where wavelengths are long relative to the lining thickness.
- Duct silencers (attenuators): Manufactured devices inserted into the ductwork containing absorptive splitters. Available in standard lengths from 600 mm to 1800 mm. A 1200 mm silencer typically provides 15-25 dB insertion loss at frequencies above 250 Hz. Critical for controlling noise from the AHU before it reaches the distribution ductwork.
- Duct routing: Every 90-degree bend provides 1-3 dB of natural attenuation. Locating plant rooms away from noise-sensitive spaces and routing ductwork through non-critical zones exploits this geometric attenuation without adding cost.
Flexible Connections
Rigid connections between rotating equipment (fans, compressors, pumps) and the building structure transmit vibration directly into the building frame, which then re-radiates as airborne noise in occupied spaces. Flexible duct connections at the fan discharge, vibration isolators under equipment, and inertia bases for large plant break this transmission path. The effect is dramatic — properly isolated equipment can reduce structure-borne noise contribution by 20-30 dB.
Diffuser Selection
Air terminal devices (diffusers and grilles) generate noise from turbulent airflow, especially when face velocities exceed design limits. A square ceiling diffuser rated NC-25 at 0.15 m/s face velocity may jump to NC-40 if the actual airflow is 30% above design because the zone has been reconfigured post-occupancy. Specifying diffusers with NC ratings 5-10 points below the room target provides margin for real-world conditions.
Equipment Selection and Location
The most effective noise control strategy is source control — selecting quieter equipment in the first place. Fan sound power data is available from every reputable manufacturer. Comparing two fans that deliver the same airflow at the same static pressure can reveal a 10-15 dB difference in radiated sound power. That 10 dB at the source is worth more than 30 dB of silencer attenuation in the ductwork, and it costs less.
Locating noisy equipment (cooling towers, chillers, generators) as far as possible from noise-sensitive spaces reduces both airborne and structure-borne transmission. Every doubling of distance provides approximately 6 dB of reduction in the direct field.
How AcousPlan Handles Noise Criteria
AcousPlan evaluates background noise against all three systems — NC, NR, and RC — in a single assessment. When you define the room type in the simulator, AcousPlan applies the appropriate target criteria based on the room's intended use and shows you which octave band is the limiting factor.
This matters because the limiting band tells you where the problem is. If a private office fails NC-30 because the 125 Hz band is 3 dB over the curve, the solution is not more ceiling tiles — it is a duct silencer or equipment isolation to address low-frequency HVAC noise. If it fails because the 4000 Hz band exceeds the curve, the problem is likely diffuser noise from excessive face velocity, which is solved by changing diffuser type or reducing airflow. AcousPlan identifies the specific octave band, so you know exactly which mitigation strategy to pursue.
The compliance report exports your measured or predicted spectrum alongside the NC, NR, and RC curves, with pass/fail status and the margin (or exceedance) at each octave band. This gives you a document you can hand to the mechanical engineer that tells them precisely what needs to change.
Start Your Noise Assessment
Ready to check whether your room meets its noise criteria target? Launch the AcousPlan simulator to model your space, set the room type, and see NC, NR, and RC compliance in seconds. No account required for your first assessment.