2 STI Past-Present-Future

After the first attempts and fruitful realization of the STI procedure the need arose to have a hand-held system for simple (person-to-person) measurements. This resulted in RASTI (Room Acoustical STI, however, also RApid STI was suggested as source of the name). RASTI became very popular and was commercially produced by B&K (see B&K Technical review, 1985, No.3). Keep in mind that a person-to-person communication is restricted to wide band and a smooth noise spectrum. RASTI was often misused for PA-system assessment with a limited bandwidth and a significant non-linear distortion (overloaded system).

In 1992 the results of a study on the revised STI were presented in a thesis (“On measuring and predicting Speech Intelligibility”) which extended the application of STI and made it more applicable in general. Also the RASTI-method was improved and replaced by STIPA (1998) to offer a wider range of communication channels. Rather than just the person-to-person application the use of band-pass limiting and non-linear distortion were included, hence PA-systems were in the picture.

The effect of masking of speech signals as originally included in the STI-procedure was extended to very high noise levels in 1999. The results (a level dependent masking correction) is given in: High-level-masking-STI-2014.

An issue studied around 2000 was the effect of non-native speech both with respect to speakers and to listeners. This resulted in a (simplified) SNR-correction of -4 dB (either for a non-native speaker, non-native listener or the combination). See STI-references-2014 no. 16 and 17.

The ultimate goal is to derive the STI from speech signals by comparison of the envelope spectra of these signals before and after transmission. This is how the original STI-concept was born. We studied this already successfully in 1975, see Houtgast and Steeneken 1975 STI-references-2014 no.10. However, the effect of non-linear distortion (related to the envelope spectrum) was not correctly accounted for. Studies are in progress to include the effect of this type of distortion on the intelligibility.

Also the relation between binaural hearing and speech intelligibility was studied in relation to a representative prediction by the STI procedure. Preliminary results are given by: STI-references-2014 no. 18.


An power-point overview (pps): Past_present_future_of_STI-2014-V1

An explanation of this presentation is given by (pdf): Past_present_future_of_STI-2014