Repository hosted by TU Delft Library

Home · Contact · About · Disclaimer ·

Prediction of speech intelligibility for public address systems in traffic tunnels

Publication files not online:

Author: Wijngaarden, S.J. van · Verhave, J.A.
Institution: TNO Defensie en Veiligheid
Source:Applied Acoustics, 4, 67, 306-323
Identifier: 15961
doi: doi:10.1016/j.apacoust.2005.07.001
Keywords: Acoustics and Audiology Traffic · Crowd evacuation · Smoke · Orientation · Safety · Runnel evacuation · Intelligibility · Prediction · STI · Tunnels · Estimation · Public address systems · Ray tracing · Reverberation · Speech transmission · Statistical methods · Intelligibility · Prediction · Speech transmission index · STI · Speech intelligibility


Traffic tunnels are generally hostile acoustic environments, both in terms of reverberation and ambient noise levels. Public address (PA) systems used to convey spoken warnings must meet stringent design requirements in order to produce sufficiently intelligible speech. To be able to predict PA system performance at tunnel design time, two di.erent speech transmission index (STI) calculation procedures were implemented. The first procedure predicts the STI based on ray-tracing simulations. Comparison with measured STI data showed that this simulation approach yields accurate intelligibility estimates. However, the procedure is time-consuming and too complex to be used by non-specialists. For this reason, a second (simpler and more efficient) procedure was developed, based on fixed non-linear regression, statistically deriving prediction functions from measured data and ray-tracing results. This procedure was compared to the approach based on ray tracing, and found to yield STI predictions closely matching those of the ray-tracing model.