Predicting the speech intelligibility in fluctuating and reverberant background noise using the Speech Transmission Index

Introduction: The concept of the Modulation Transfer Function (MTF) in the field of room acoustics was introduced by Houtgast & Steeneken (1973). The MTF can be used to predict speech intelligibility in stationary noise and reverberation and can be expressed in one single value: the Speech Transmission Index (STI). The drawback of STI meas-urements using the classical, direct method that it is time-consuming and that it is not val-idated for fluctuating background noise (IEC60268-16, 2011). However, the MTF as a result of reverberation can also be calculated ―indirectly‖ using impulse response meas-urements (based on Schroeder, 1978). Furthermore, when calculating the STI per time segment and applying the same time averaging as used by Rhebergen et al., 2006 (regard-ing the Extended Speech Intelligibility Index), the ―Extended‖ Speech Transmission In-dex (ESTI) can be calculated in fluctuating noise. The aim of the current research is to obtain validation data using speech intelligibility experiments. Methods: Monaural speech intelligibility experiments using headphones were conducted. The 50% Speech Reception Threshold (SRT, Plomp & Mimpen, 1979) was measured us-ing sentences (Versfeld et al. 2000) and digit triplets (Smits et al. 2013). Spectrally matched stationary noise and 8 Hz interrupted noise were used as background noise. The SRTs were measured with reverberation times between 0 and 1.2 seconds. In the first phase of the experiments we measured SRTs in normal hearing listeners (work in pro-gress). Results: In stationary noise the effects of reverberation on the ESTI are small, which is in agreement with the results of Duquesnoy & Plomp (1980). For 8Hz interrupted noise, the ESTI values in the conditions without reverberation are somewhat lower, but for high re-verberation times (T60 = 1.2 s) the ESTI values approach those for stationary noise. The-se main effects are found both for sentences and for digit triplets. Discussion and Conclusion: Due to the gaps in the noise the SRT in interrupted noise is lower than in stationary noise, but this advantage disappears when adding reverberation. This effect is reflected in the ESTI values. Our preliminary results present possibilities to examine the ESTI model, but more experiments will be done in the near future in order to validate the model approach. We also aim to use a wider variety of background noises and to include hearing impaired subjects and children. A major goal is to use the ESTI regularly in the evaluation of listening conditions in classrooms