The STI can be calculated for conditions where speech is masked by a noise signal. Hence, for prediction purposes, no measurement is required if the speech level and noise spectrum are known and no other disturbances are concerned (i.e. reverberation, echoes or non-linear distortion, for band-pass limiting some knowledge is required in order to convert the speech spectrum to the actual transfer function).
Another application is to use measurement results from a real measurements. A STIPA measuring device may deliver these data. It opens the possibility to predict the effect on the STI by changing speech or noise levels.
The spreadsheet given below uses as input: -the noise spectrum (octave band levels from 125Hz up to 8kHz), -the speaking distance from the listener or microphone (a normalized speech level, representing the vocal effort, is assumed). It calculates the STI for both, male and female speakers.
This spreadsheet is developed by: © Embedded Acoustics:
STI calculation spreadsheet (xls): STICalc_SM50_v1