Skip to content
TechnoGuru — Think Technology, Think TechnoGuru

/ ELV

Voice-evacuation intelligibility: STI per seat in a real evacuation, not a clean room

By Pranab Kumar BeriyaFounder & Chief Executive Officer·Published 21 May 2026·7 minute read·ELV

Quick answer

Voice-evacuation PA per IS 16102 / IEC 60849 / EN 54-16 must deliver STI ≥ 0.50 at every seat during an actual evacuation — not just at commissioning. The gap between the empty-room measurement and the real evacuation event is closed by zone-loudness margin against measured ambient noise (≥ 10 dB above evac-event ambient, not standing ambient), multi-language message ordering, message timing against the alarm-sounder discipline, hardware-enforced priority over routine paging, supervised cabling with per-zone fault reporting, battery-backed amplification for 30 minutes minimum, and re-measurement after any room-acoustic change (carpet refresh, ceiling re-treatment, layout reconfiguration).

Voice-evacuation is the most consequence-bound PA system in any building. The acceptance criterion is intelligibility during an evacuation event — when the alarm is sounding, the doors are opening, the population is in motion, and the announcement has to compete against everything else for cognitive bandwidth. STI ≥ 0.50 at every seat in the empty room is the floor; STI ≥ 0.50 during the actual evacuation is the engineering target.

## Zone-loudness margin against evacuation-event ambient

The catalogue zone-loudness specification is typically ≥ 10 dB above standing ambient (the room's background noise floor in normal use). The evacuation-event ambient is much higher — alarm sounders at 80-90 dBA, audience movement noise at 65-70 dBA, door operation noise at 60-65 dBA. The voice-evacuation PA must deliver ≥ 10 dB above the evacuation-event ambient, which means 95-105 dBA at every seat under evacuation conditions. Specifying against standing ambient and discovering the gap during the event is operationally late.

## Multi-language message ordering

Public-occupancy buildings in India routinely require evacuation announcements in 2-4 languages (English, Hindi, regional language, sometimes additional for transit hubs and airports). The ordering and timing of the languages is part of the engineering: each language takes 12-18 seconds for a complete evacuation message; the message library must be sequenced so the most-comprehended language plays first, followed by additional languages on a rotating cycle, with sufficient pause for the population to act on the message rather than waiting for the next language. Generic preset libraries miss this.

## Hardware-enforced priority over routine paging

PA voice-evacuation priority must be hardware-enforced, not software-configured. The fire-alarm panel's voice-evac module wires into the highest-priority input on every zone amplifier; on a fire-alarm trigger, the zone amplifiers switch to the voice-evac source regardless of what the routine PA is broadcasting. A software-configured priority that a routine paging operation can over-ride is a procedural failure waiting to happen — and we test the hardware priority on every commissioning by triggering a fire-alarm event during a routine page.

## Supervised cabling with per-zone fault reporting

IS 16102 / EN 54-16 voice-evacuation requires supervised cabling — the system monitors every cable run between the panel and the amplifier, between the amplifier and every speaker, with per-zone fault reporting on cable open, cable short or amplifier failure. Supervision is engineered at the cable layer with end-of-line resistors or active line-monitoring; without it, a single cable fault collapses a whole zone silently and the operator discovers the fault during the evacuation event.

## Battery-backed amplification for 30 minutes minimum

Voice-evacuation amplification must continue through power failure — the standard requires 30 minutes minimum battery autonomy under full-load voice-evacuation discharge. The battery bank is sized against the full-discharge load profile, not the standing-charge load; sealed maintenance-free batteries on a 3-year refresh cycle for the 30-minute guarantee.

## Re-measurement after any room-acoustic change

STI is a function of the room's acoustic envelope — carpet, ceiling treatment, audience-chair upholstery, layout reconfiguration all shift the RT60 and therefore the achievable STI. Any room-acoustic change triggers an STI re-measurement on the AMC schedule; the original commissioning STI is the starting baseline, not the lifetime guarantee.

## Callout — what evacuation-PA procurement most miss

**STI per seat at commissioning is the starting point, not the operational guarantee.** The evacuation-event STI is what decides whether the population evacuates safely; engineer the zone-loudness margin, the multi-language sequencing, the hardware priority, the supervised cabling, the battery backup and the AMC re-measurement discipline against the event, not the clean-room test.

## Reference deployment context

Capital Cultural Hall Kohima (1,800 seats) and Town Hall Auditorium Dimapur (374 seats) both run voice-evacuation overlay on the PA chain per IS 16102, with hardware-enforced priority on the zone amplifier, supervised cabling per zone, battery-backed amplification for ≥ 30 minutes and per-zone fault reporting on the fire-alarm panel.

## References

1. IS 16102:2014 — Voice Alarm Systems — Sound systems for emergency purposes.

2. IEC 60849 — Sound systems for emergency purposes (international reference).

3. EN 54-16 — Fire detection and fire alarm systems — Voice alarm control and indicating equipment.

4. NBC 2016 Part 4 — fire and life-safety provisions for assembly occupancy.

Auditorium signal architecture with voice-evacuation overlay

auditorium-signal-architecture
Civic auditorium signal architectureA civic / cultural auditorium AV signal architecture. Inputs (gooseneck and lavalier microphones, line-in sources, HDMI presentation, livestream feed) route through a digital mixer to a DSP with feedback suppression, EQ and named-preset library (Civic, Cultural, Ceremonial). DSP feeds line-array audio output, stage monitors, LED-cluster video wall, DMX stage lighting and a recording / livestream encoder. Auditorium-staff console recalls a preset and orchestrates all five output classes from a single gesture.Auditorium signal architecture · named-preset libraryCivic · Cultural · Ceremonial modes recallable by in-house staff · DSP preset is the operational contractSourcesMic · line · HDMI · streamLectern gooseneck × 2Presidium gooseneck × 6Wireless lav × 8Choir hanging × 2HDMI presentationLivestream feed inPre-recorded EVACMixer + consoleDigital · scene memorySoundcraft Si ExpressionMini Stage Box × 1Wireless rx rack × 8Scene memory (16 banks)Operator faders + faderbankTalkback intercomPre-flight checklistDSP + processingBSS BLU · feedback suppressBSS BLU 50 / BLU 100Auto Feedback SuppressorCivic presetCultural presetCeremonial presetEQ shelf + parametricDelay + zone matrixOutputsAudio · video · lighting · recordJBL SRX line arraySRX928S subwoofers × 4Stage monitor wedgesCentre LED cluster (P2.5)DMX stage lightingRecording + livestreamFoyer + lobby BGMAuditorium-staff console recalls a preset; the preset coordinates DSP profile, line-array gain shading, LED-cluster brightness profile and DMX scenePre-flight checklist locked to operator scene memory · mic count verified per session · DSP preset re-tested after every firmware refreshSTI measurement at commissioning per AES-15id · re-measured after any room-acoustic change · 374-seat venues → panel-and-cloud; above 800 → fluid treatment
DSP-driven signal architecture with named-preset library; the fire-alarm panel's voice-evac module wires into the highest-priority input on every zone amplifier for hardware-enforced override.

Key engineering takeaways

  1. Voice-evacuation STI must hold at every seat during the actual evacuation event — not just at commissioning.
  2. Zone-loudness margin is engineered against evacuation-event ambient (alarms + movement + doors), not standing ambient.
  3. Multi-language message ordering and timing is part of the engineering — generic preset libraries miss this.
  4. Hardware-enforced priority over routine paging — software-configured priority is a procedural failure waiting to happen.
  5. Supervised cabling with per-zone fault reporting per IS 16102 / EN 54-16 — a single cable fault otherwise collapses a whole zone silently.
  6. Battery-backed amplification for 30 minutes minimum at full-load voice-evacuation discharge — 3-year sealed-MF battery refresh cycle.
  7. STI re-measurement after any room-acoustic change — carpet refresh, ceiling re-treatment, layout reconfiguration all shift the achievable STI.

/ Frequently asked

Quick answers from the practice.

What STI target does IS 16102 require?
STI ≥ 0.50 at every position in the served area for emergency-use voice alarm. We typically design and commission against STI ≥ 0.55 to preserve margin against room-acoustic drift over the AMC lifecycle.
How is multi-language sequencing engineered?
Message library sequenced so the most-comprehended language plays first, followed by additional languages on a rotating cycle, with sufficient pause (typically 4-6 seconds) for population to act on the message before the next language begins. Sequencing is part of the commissioning deliverable, signed off by the building operator.
Will TechnoGuru deliver the IS 16102 compliance pack?
Yes — measurement report (STI per seat, SPL per seat, calibrated source), zone-and-amplifier discipline, pre-emption test record, signed compliance against the standard. Standard inclusion on any building requiring a Fire NOC.

/ What to do next

Three next steps for voice-evacuation scope

/ About the author

Pranab Kumar Beriya Founder & Chief Executive Officer

Founder of TechnoGuru; sixteen years of practice in residential cinema, automation and turnkey systems integration across eastern India and the wider sub-continent. AVIXA Certified, K-Array Designer, CEDIA Member, HAA Level 1 Calibrator, Rako-DALI trained, AMX-certified, Harman BSS programming-certified, Alcatel-Lucent OXO Connect-certified.

/ Discuss your project

If this article matches a brief you are working on, the next step is a thirty-minute call with a project lead.

We do not run sales pipelines. The first reply comes from a project lead, within two working days, and it goes straight to the engineering question rather than a brochure.

Begin a brief
Voice-evacuation intelligibility: STI per seat in a real evacuation, not a clean room | TechnoGuru