IOA Diploma in Acoustics and Noise Control — Bangkok, from 16 Nov 2026
← Our Services
Our Services

PAGA and voice alarm testing — STIPA speech intelligibility and alarm audibility measured live, indoors and outdoors

A public address and general alarm (PAGA) system, a voice evacuation system or a fire alarm passes or fails on two questions: can the message be heard above the noise, and can it be understood. Geonoise measures both on site. Speech intelligibility is measured as STI with the STIPA method of IEC 60268-16, and audibility as the alarm level and its margin above the ambient noise. We measure zone by zone, indoors and outdoors, with the system running live. Each position gives a result within about half a minute, so the integrator can adjust level, equalisation or delay during the walk-through and we re-measure on the spot.

Geonoise is an independent consultancy. We do not supply, install or maintain loudspeakers, amplifiers or fire alarm panels, and we do not recommend suppliers. This is why owners, fire engineers, insurers and EPC contractors appoint us as the commissioning witness: the report states what the system achieves, not what anyone hoped it would. All measurements use Class 1 instruments calibrated by an ISO/IEC 17025 accredited laboratory. For the background to the metric, read STI and STIPA explained. To check intelligibility before anything is installed, see speech intelligibility prediction.

Systems we test

  • PAGA in refineries, petrochemical and chemical plants, power plants, ports and jetties: process areas, compressor and pump houses, control rooms, muster points
  • Voice alarm and voice evacuation (EVAC) systems linked to the fire alarm in high-rise offices, hotels, hospitals, shopping centres and condominiums
  • Fire alarm sounders, bells and alarm tones: audibility in every occupied area, including bedrooms where the specification requires a level at the bedhead
  • Transit: airport terminals and piers, rail and MRT stations, platforms, concourses, tunnels and bus terminals
  • Factories and warehouses with high ambient noise and large reverberant halls, where a PA that is loud enough is often still not intelligible
  • Stadiums, arenas, exhibition and convention centres, campuses and outdoor plazas

What we measure

  • Speech intelligibility as STI, measured with the STIPA method of IEC 60268-16 at a grid of listener positions in each zone, reported per position and as zone mean and minimum
  • Alarm audibility: the A-weighted and octave-band level of the alarm tone or message at listener height, and its margin above the ambient noise
  • Ambient noise in each zone with the plant, traffic or crowd as in normal operation: Leq and the octave-band spectrum. Where the system cannot be tested at full operating noise, this spectrum is used to correct the STIPA result as IEC 60268-16 allows
  • Reverberation time to ISO 3382 in atria, halls and terminals, where it is usually what limits intelligibility
  • Coverage: how much the level of the message varies across a zone, and which positions fall outside it
  • Outdoors: long-delay echoes between distributed horn loudspeakers, coverage at distance, and the wind and weather at the time of the test

How live testing works

Most zones are completed in one site visit, with the integrator present and the system under its own control.

  • Before the visit we agree the zones, the listener positions on plan, the criteria and which specification governs, so the result cannot be disputed afterwards
  • The system plays the IEC 60268-16 STIPA test signal at the level of the real emergency message, from our calibrated signal source or loaded into the system's own message store
  • Each position is measured for 15 to 25 seconds, at least twice, and a third time when the two readings differ by more than 0.03
  • A position that fails is flagged on the spot. The integrator adjusts level, equalisation, delay or loudspeaker tap, and we re-measure and record before and after
  • Where the plant cannot run during the test, the ambient noise is measured separately during normal operation and combined with the quiet-condition STIPA result, which gives the intelligibility the listener will actually experience
  • Night or weekend testing for hospitals, airports and operating plants that cannot sound alarms during working hours

Acceptance criteria

The criterion comes from your project specification, and we report against the one it names. Where the specification is silent, we agree the criterion with the owner before the test, not after it. As a guide to what specifications commonly require:

  • Voice alarm intelligibility: commonly a mean STI of at least 0.50, with no individual position below about 0.45. ISO 7240-19, CEN/TS 54-32 and BS 5839-8 all use criteria of this order
  • NFPA 72 (Annex D) is used on American-specified projects and sets out its own intelligibility test procedure and acceptance rules
  • Alarm audibility: usually a minimum level and a margin above the ambient noise. BS 5839-1 commonly applies 65 dB(A) in general areas and 75 dB(A) at the bedhead where people sleep; NFPA 72 public mode uses 15 dB above the average ambient level. Margins of 10 to 15 dB are typical, depending on the standard
  • Oil and gas and marine PAGA specifications often require both an STI and a margin above ambient in noisy areas, and visual beacons where the noise makes an audible alarm impractical
  • Projects in Thailand: the fire alarm standard of the Engineering Institute of Thailand (EIT) and the project specification apply

Why systems fail, and what fixes them

A failing zone almost always has one of a few causes, and the measurement shows which. We state the cause and the measure, with a fallback. Your integrator or designer implements it; we re-test.

  • Too few loudspeakers in a reverberant space: the level is fine but STI is low. Fix: more loudspeakers at lower level, closer to listeners, or directional (column) loudspeakers. Fallback: absorption in the space
  • Level set too high to overcome noise: the room reverberates and the loudspeakers distort, so STI falls as level rises. Fix: lower level with better coverage
  • Plant noise masking the message: Fix: local loudspeakers near the noise source, a higher-level zone, or visual alarms where speech cannot work
  • Long-delay echoes outdoors between widely spaced horns: Fix: delay alignment or re-aiming. Fallback: shorter horn spacing
  • Poor equalisation or a poorly recorded message: Fix: equalise to the space and re-record the message at a clear, measured pace

What you receive

  • A measurement report in Thai or English with the standard, the test signal and level, the instruments and calibration, and the positions marked on plan
  • STI per position with the zone mean and minimum, and audibility levels and margins against ambient, compared against the criterion with a clear pass or fail per zone
  • Ambient noise spectra per zone, and the noise correction applied where the test was made in quiet conditions
  • Before-and-after records for every adjustment made during the visit
  • For each failing position: the likely cause, the recommended measure and a fallback, and a re-test once the work is done

Standards & method

IEC 60268-16:2020 (STI, STIPA) · ISO 7240-19 (voice alarm design and commissioning) · CEN/TS 54-32 · BS 5839-8 (voice alarm) · BS 5839-1 (fire detection and alarm) · NFPA 72 Annex D (intelligibility) · ISO 3382-1/-2 (reverberation time) · IEC 61672-1 Class 1 · instruments calibrated by an ISO/IEC 17025 accredited laboratory

Related services

This work is usually combined with Speech Intelligibility Prediction (STI) and Industrial & Occupational. For the background before you decide, read STI and STIPA: how speech intelligibility is measured for PA, voice alarm and PAGA systems in Noise 101.

FAQ

What is STIPA?

STIPA is the simplified method of measuring the Speech Transmission Index (STI) in IEC 60268-16. A special test signal is played through the system and a meter measures how much of its modulation survives at the listener, in seven octave bands. The result is a number from 0 to 1, and a measurement takes 15 to 25 seconds, which makes it practical for testing a PA or voice alarm system position by position.

What STI does a voice alarm or PAGA system need?

Specifications commonly require a mean STI of at least 0.50 in each zone, with no position below about 0.45. Your project specification governs, and we report against the figure it names.

Can you test while the building or plant is operating?

Yes. We can measure with the plant running and the alarm sounding, at night or at weekends where needed. Where that is not possible, we measure the ambient noise separately during normal operation and correct the STIPA result for it, as IEC 60268-16 allows.

Do you test outdoors?

Yes. Process areas, jetties, tank farms, car parks, stadiums and plazas. Outdoors the usual problems are coverage at distance and long-delay echoes between horns. We record wind and weather with the result, because both affect it.

Do you test fire alarm sounders as well as voice messages?

Yes. For sounders, bells and alarm tones the question is audibility: the level at listener height in every occupied area and its margin above the ambient noise. Intelligibility applies only where the system speaks.

Do you also supply or repair the system?

No. Geonoise is independent and does not sell, install or maintain any equipment, and we do not recommend suppliers. The report gives the cause of each failure and the measure needed. Your integrator carries it out, and we re-test.

How long does a test take?

A zone of 10 to 20 positions usually takes one to two hours including set-up and repeats. A whole building or plant is planned zone by zone, and we give you the schedule with the quotation.

Scope your project →Free Consultation →