Emergency communication systems are a critical layer of life‑safety strategy. They provide warning when a hazardous event is occurring and also direct occupants to appropriate actions.
The development of in-building voice alarm systems has two distinct origins. EVACS requirements grew from catastrophic high-rise fires, most notably the 1980 MGM Grand Hotel fire, where tone-only alarms failed to direct occupants, one of the factors contributing to the death of 85 people, leading leading NFPA 101 to mandate intelligible voice systems in high-rise and large assembly occupancies.
Mass Notification System (MNS) requirements followed a different path: the 1996 Khobar Towers bombing exposed the absence of effective mass warning systems in military facilities, prompting US Department of Defense (DoD) to mandate MNS through its Unified Facilities Criteria, and subsequently leading NFPA to integrate MNS guidance into the 2007 edition of NFPA 72 and full enforceable requirements into the 2010 edition under the new Chapter 24, ‘Emergency Communication Systems’ (ECS).
System designers and fire protection engineers must understand the difference between making a message loud enough to hear (audibility) and making it understandable (intelligibility), recognising when voice messages are required without over‑designing.
Audibility and its code requirements
Audibility means that alarm signals are perceptible by the normal human ear and the code sets quantitative thresholds for that. NFPA 72 requires audible notification appliances to deliver sound levels at least 15 dB above the average ambient sound or 5 dB above the maximum ambient sound lasting at least 60 seconds, whichever is greater. For public‑mode signalling, this ensures messages can be clearly heard by all occupants.
In sleeping areas, the requirement is more stringent: the sound level at the pillow must be 15 dB above average ambient, 5 dB above the maximum sound of at least 60 seconds, or 75 dBA whichever is greater. From a design perspective, these levels translate into two practical rules:
- Measure ambient noise correctly. NFPA 72 directs designers to document the greatest expected average ambient vs the expected maximum sound level lasting 60 seconds and to include these values in design documents. Annex A provides typical ambient sound levels by occupancy: business occupancies (≈55 dBA), educational (45 dBA), industrial (80 dBA), mercantile (40 dBA) and residential (35 dBA).
- Avoid over‑amplifying speakers. Overdriving a speaker to meet a higher dBA will result in distorting the voice signals; increasing the number of speakers while using lower power taps is usually a better option.
Because of these requirements the audibility is enforceable and placed in the body of the code. Intelligibility in contrast remains largely in the annex sections as guidance.
Intelligibility: Concept and code guidance
Intelligibility refers to the level to which a spoken message is understandable. It is influenced by acoustics, signal‑to‑noise ratio, distortion, frequency response and reverberation. NFPA 72 defines an intelligiblemessage as one that is capable of being understood, clear and comprehensible by the occupants.
Why intelligibility matters
Before 1999, NFPA 72 focused on audibility only. Starting with the 1999 edition, it also required that voice messages be intelligible. Designers and AHJs struggled because no specific quantitative criteria were provided back then. The 2010 code edition made intelligibility an explicit requirement under Chapter 24 and added Annex D on speech intelligibility as explanatory material. NFPA 72 (2013 and later) emphasised that voice messages within acoustically distinguishable spaces (ADS) where intelligibility is required must be intelligible. However, NFPA 72 also states that quantitative measurements of intelligibility (e.g., STI or CIS) are not mandatory; a simple listen test is acceptable. Measurement tools such as Speech Transmission Index for Public Address (STIPA) can be used when desired but are not as a requirement unless another code or owner specification required it.
Factors affecting intelligibility
Eaton’s white paper Designing for Intelligibility vs. Audibility provides practical insight. It explains that the signal‑to‑noise ratio (SNR – the difference between the speaker output and ambient noise) is critical; raising volume above 15 dB over ambient improves intelligibility but yields diminishing returns beyond that. Harmonic distortion, frequency response (speech energy lies between 500 Hz and 4 kHz) and reverberation also influence comprehension.
Speech transmission metrics
Annex D discusses objective methods such as Common Intelligibility Scale (CIS) and Speech Transmission Index (STI). A typical criteria is 0.70 CIS (≈0.50 STI), however NFPA 72 does not enforce these values. STIPA testing uses a talk‑box to inject test signals and measures the resulting intelligibility score. Subjective testing, using trained listeners, is also permitted and possible.
Acoustically Distinguishable Spaces (ADS)
An important concept of NFPA 72 is the concept of the acoustically distinguishable space which is a portion of an emergency communication notification zone distinguished from adjacent spaces by different acoustical, environmental or use characteristics.
Examples include rooms separated by changes in ceiling height, or an open office against a reverberant atrium. The system designer must identify all ADS during the planning phase and assign each ADS as requiring or not requiring intelligibility.
Identifying ADS and deciding where intelligibility is required
NFPA 72 Annex A recommends that a 20% change in ceiling elevation or a major change in acoustic finishes typically signals a new ADS.
ADS classification has to consider:
- Use and occupancy: Spaces where occupants need verbal instructions (classrooms, assembly areas, public corridors) usually require intelligible voice messages. Private offices, utility rooms or storage spaces may not.
- Acoustic properties: Areas with high reverberation (hard surfaces, large volumes) are more challenging. An informative presentation prepared by Wayne D. Moore with AFAA titled “Understanding the Why Behind NFPA 72 Requirements” notes that non‑challenging ADSs such as traditional offices, hotel guestrooms, carpeted spaces often achieve intelligibility simply by meeting audibility requirements.
Conversely, atriums, reception lobbies with reflective finishes or high ceilings require closer speaker spacing and lower tap settings. In ADS with ambient noise above 85 dBA, intelligibility may not be attainable, and an alternate notification method must be utilised. - Background noise: Identify sources such as HVAC equipment, industrial machinery or music. Moore’s presentation lists average ambient levels by occupancy (e.g., 55 dBA for business, 80 dBA for industrial). ADS with highly variable noise may require additional speakers or sound masking management in place.
Designing within an ADS
For standard spaces with 2.5 to 3.5 m (Approx 8 to 12 ft) ceilings, drop‑in acoustical tiles and average furniture, NFPA 72 handbook offers simple guidelines as follows: install ceiling‑mounted speakers in all normally occupiable spaces and corridors spaced no more than twice the ceiling height, or use a modelling program. The speaker tap setting should be selected to meet audibility without distortion; if audibility cannot be achieved at a lower tap, adding more speakers instead of increasing wattage would provide better results. Moore’s presentation emphasises that higher taps applied to achieve audibility can distort intelligibility.
For acoustically challenging ADS, consider:
- Speaker spacing: Closer spacing reduces reflections and increases direct‑to‑reverberant energy.
- Speaker type and frequency response: Use wide‑bandwidth speakers (400 Hz to 4 kHz for fire alarm) and high‑fidelity models when necessary.
- Acoustic treatments: Where possible, add absorptive finishes such as carpets, acoustic panels or soft furnishings to reduce reverberation and improve SNR. Normal building materials with acoustic properties can mitigate average reverberation.
Practical design example: Office floor with atrium
Consider a mixed‑use building with a 3.0 m (10 ft) high open office area, adjacent to a double‑height atrium, meeting rooms and a mechanical plant. Ambient measurements taken during typical occupancy showed 55 dBA in the office, 65 dBA in the atrium (due to a fountain), 50 dBA in meeting rooms, and 85 dBA in the mechanical plant. Using NFPA 72 audibility requirements:
- Office open area: Target sound level = 55 dBA + 15 dB = 70 dBA.
Use distributed ceiling speakers spaced less than 6 metres (twice the 3m ceiling height). A design using one speaker per 40 m² at 1 W tap provides ~75 dBA at 3m and meets the requirement after accounting for 6 dB drop per doubling of distance (100 dB at 3 m → 94 dB at 6m, etc.).
Intelligibility is expected to be acceptable because the space is non‑challenging; no quantitative measurement is required.
- Atrium: Target sound level = 65 dBA + 15 dB = 80 dBA.
The double‑height, reflective finishes create noticeable reverberation. Speaker spacing should be closer, perhaps 4.5m, using high‑fidelity speakers with wide bandwidth. (Note that the AHJ may require STIPA measurements). If STI results fall below 0.50 threshold, consider adding more sound absorptive materials or supplemental speakers (less spacing). - Meeting rooms: Each room has its own ADS; design for 50 dBA + 15 dB = 65 dBA at the user location. A single ceiling speaker will usually suffice. It is recommended to test intelligibility subjectively.
- Mechanical plant: Ambient noise exceeds 85 dBA; NFPA 72 acknowledges that intelligibility might not be attainable and an alternate means of notification (e.g., strobe lights, low‑frequency alarms) is required. This is commonly the case where strobes are used in addition to the EVAC speakers’ layer.
Interfacing EVACS with the fire alarm system
EVACS must be fully integrated with the fire alarm control unit, with any initiating device automatically triggering the appropriate voice message to the correct notification zones while simultaneously activating emergency control functions such as elevator recall, HVAC shutdown and door holder release.
This integrated approach is the industry standard for buildings where voice notification is required, and one of the most direct cost benefits is device consolidation. Speaker-strobes replace both conventional horns and discrete strobes in a single unit, reducing field devices, wiring and installation labour. Note that once EVACS is adopted in a zone, tone-only horns must be fully replaced by speaker, as mixing the two appliance types within the same coverage area compromises voice intelligibility.
Beyond device savings, pathway survivability requirements represent the largest wiring cost variable. Systems using staged or partial evacuation require level 2 or 3 survivability throughout the riser, whereas a total evacuation strategy permits level 0 or 1, significantly reducing wiring costs.
Testing and acceptance
During acceptance testing, verify that the measured sound levels at listeners points in each ADS match the documented design levels. Use a calibrated sound level meter at 2.7m above the floor as NFPA 72 requires. If the AHJ calls for an intelligibility test, perform either a subjective test (trained listeners who judge clarity) or an objective test using a calibrated STIPA equipment. Remember that intelligibility testing is intended to identify issues to remedy; it is not necessarily a pass/fail test. If the results are marginal, explore adjustments such as re‑balancing amplifier gains, re‑tapping speakers or adding absorptive materials and furniture.
Engineering judgment
NFPA 72 provides a framework but leaves room for professional judgment. System designers should: (1) identify occupant notification goals, (2) evaluate each space’s acoustics and ambient noise, (3) select equipment and layout to meet audibility and, where required, intelligibility, and (4) document and verify performance.
By understanding the distinction between audibility and intelligibility, correctly assigning ADS, and applying acoustic principles, engineers can design emergency voice alarm systems that meet code requirements, avoid over‑design, and most importantly, deliver clear instructions when they are needed most.
