Search

Free tool

What STI will we get?

The Speech Transmission Index (STI) rates how much of the rise and fall of speech survives the trip from the loudspeaker to the listener, on a scale from 0 to 1. Before anything is installed, the honest answer to "what STI will we get?" is a range. The room sets the ceiling, and noise and speech level decide how close you get to it. Enter what you know and the tool shows that range. Everything runs in your browser, and the result needs no email.

The tool

Reverberation time
Speech above the noise
Listener and system
Meter reading consistency

Even with nothing changing in the room, repeat STIPA readings differ by about 0.02 to 0.03. Set to zero to see the calculation alone.

Estimated STI

What was entered

    An estimate to plan with, not a measurement or a compliance figure. Worked out at proaudium.com/tools/sti-estimator/ on .

    Band by band

    Keep a copy

    Get this estimate by email, to keep or to pass on to a project manager.

    This sends one email, and nothing else unless you tick a box. If you ask us to look at the figures, we reply to the same address. How your email is used.

    What decides the STI

    Three things, in this order. First the room: its reverberation time, how long a sound takes to die away, blurs the quick rises and falls that make words distinct, and no loudspeaker can put back what the room takes away. Second the noise: background noise fills in the quiet moments between syllables. Third the speech level, which helps against noise but only up to the limit the room allows. The tool shows that limit as "Best this room allows with no noise".

    Above about 1 s of reverberation in the speech frequencies, clear speech becomes harder to achieve, and the tool marks it in yellow. Above about 2 s, it marks it in red: clear announcements are then hard to achieve with any system without careful loudspeaker choice and, usually, acoustic treatment.

    Reading the result

    The range is one standard deviation either side of the middle figure, so about two times in three a reading in the finished space would land inside it. "Chance of reading 0.50 or more" is the share of the 2000 calculations that reached 0.50. A range that straddles 0.50 means the outcome is genuinely uncertain, and a single predicted figure would hide that.

    How the estimate is worked out

    For each of seven octave bands from 125 Hz to 8 kHz, and each of 14 rates at which speech rises and falls, the tool works out how much of that rise and fall survives the reverberation and the noise. It follows the method in BS EN IEC 60268-16:2020, Annex A: the band weightings, the masking of one band by the band below it, the threshold of hearing and the test-signal spectrum. It then repeats the calculation 2000 times with the reverberation time and the speech above the noise varied by the amounts you set, and adds the variation between repeat meter readings. Repeat STIPA readings in an unchanged room typically differ by 0.02 to 0.03 (Mapp, 2006).

    With "Add the room information", each band's reverberation time comes from Sabine's formula, with Eyring's shown alongside because Sabine overstates the time in a well-damped room. The absorption figures are generic ones from a published table (sengpielaudio.com); a manufacturer's tested figures for the actual product are better. Air absorption follows ISO 9613-1 at 20 °C and 50 % relative humidity.

    What it can't tell you

    It assumes steady noise, one evenly spread reverberant sound field and one listener with normal hearing. It can't see echoes, distortion, signal processing, the talker or the listener. Treat the result as an estimate to plan with, not as a measurement or a compliance figure. A real space needs measuring once it is built: consultancy, acoustic modelling and specification.