Speech intelligibility prediction — STI calculated at design stage for PA, voice alarm and PAGA systems
Fixing a PA or voice alarm system after it is installed means more loudspeakers, more cable and work inside finished ceilings, usually against a handover date. Predicting STI at design stage, from the room geometry, the surface finishes, the loudspeaker layout and the expected noise, shows which zones will fail while they are still a drawing. Geonoise builds a 3D acoustic model of the spaces that matter, calculates STI and level maps over the listener area, and checks the result against the criterion in your specification.
We check the design; we do not supply it. Geonoise is an independent consultancy and does not sell loudspeakers or recommend a brand, so the review tells you whether the layout your designer or integrator proposed will meet the specification, and if not, what has to change. The same model answers the question that is usually argued without numbers: more loudspeakers, or acoustic treatment. After installation, the prediction is confirmed by measurement; see PAGA and voice alarm STIPA testing. For the metric itself, read STI and STIPA explained.
When a prediction pays for itself
- Design review or tender evaluation of a PAGA system for a new plant, terminal or station, before the equipment is ordered
- Atria, concourses, halls and stadiums with long reverberation, where a layout that looks adequate on plan often fails
- An existing system that failed its commissioning test: the model shows which remedy works before anything is changed on site
- Checking a contractor's intelligibility claim, which is often a single figure without the assumptions behind it
- Deciding between more loudspeakers and acoustic treatment, with the effect of each calculated rather than argued
What we calculate
- STI maps over the listener plane in each zone, with and without the expected ambient noise
- Direct and total sound level from the loudspeakers, and how evenly it covers the zone
- Reverberation time and early decay from the finishes, and how much each proposed treatment changes them
- The effect of ambient noise, from equipment noise data or from measurement in an existing facility
- Outdoors: coverage at distance from horn arrays, long-delay echoes between them, and the delay settings that remove them
- Sensitivity: how the result changes with loudspeaker spacing, aiming, level and absorption, so the design has a margin and not just a pass
How the work is done
- Inputs: drawings and sections, the finishes schedule, the proposed loudspeaker layout and types with the manufacturer's data in a standard format, and ambient noise data
- In an existing building, we measure reverberation time and ambient noise first and calibrate the model against them, which removes most of the uncertainty in absorption data
- We calculate, review the results with the designer, and iterate on the layout until each zone meets the criterion with a margin
- The assumptions are stated in the report. A prediction is only as good as the absorption and loudspeaker data behind it, which is why we always recommend a STIPA test after installation
What you receive
- A report in Thai or English with STI and level maps per zone, and a table of zone mean and minimum STI against the criterion
- The zones at risk, the design changes that bring them into compliance, and a fallback measure for each
- The assumptions, input data and model limits, set out so another engineer can check them
- A test plan for verification after installation: zones, positions and criteria, ready to use for commissioning
Standards & method
IEC 60268-16:2020 (STI) · ISO 7240-19 · CEN/TS 54-32 · BS 5839-8 · NFPA 72 Annex D · ISO 3382-1/-2 (room acoustic parameters) · ISO 9613-2 (outdoor sound propagation)
Related services
This work is usually combined with PAGA & Voice Alarm Testing (STIPA) and AV & Speech Intelligibility. 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
Can STI be calculated before a system is installed?
Yes. With a 3D model of the space, the absorption of the finishes, the loudspeaker data and the expected noise, STI can be calculated at every listener position. It is the most reliable way to find a zone that will fail while the fix is still cheap.
How accurate is a prediction?
It depends on the input data, mainly the absorption of the finishes and the loudspeaker data. In an existing building we calibrate the model against measured reverberation time, which removes most of the uncertainty. We state the assumptions and always recommend a STIPA verification after installation.
Will you design the PA system?
We check and improve the layout against the criterion, and say what has to change. We do not specify a brand or supply equipment; the system design and supply stay with your designer or integrator. That keeps the review independent.
More loudspeakers or acoustic treatment: which is better?
It depends on the space. In a very reverberant hall, absorption often raises STI more than extra loudspeakers do; in a large, moderately reverberant space, more loudspeakers closer to the listener usually win. The model calculates both, so the choice rests on numbers.
Do you also predict outdoor PAGA coverage?
Yes. For plants, ports and campuses we calculate coverage from the horn positions, the long-delay echoes between them and the ambient noise in each area, and give delay settings and positions that keep the message intelligible.