Hi Guys, I have been asked this question but have not been able to find the answer. Could you please take a look and let me know if you can help? Question:

This relates to the speech presentation levels for the STI calcs with a background noise spectrum. What is or how do I find the speech presentation sound pressure level? The SPL maps, from what I gather, are the pink noise response of the speaker system. The STI results change a function of speaker gain etc, but I don’t know how to find the speech SPL which has a shaped spectrum compared to the pink noise input.There must be a simple dB correction that is being applied in the software to convert the pink noise response of the speaker to the STI speech spectrum. The STI speech spectrum, unlike pink noise, is not flat so the SPL will be different for pink noise and speech using the same speaker gain.

It is the speaker response to the STI spectrum (SPL with the speech spectrum) that I need so I can design for a minimum speech intelligibility while also ensuring I do not exceed environmental noise limits.

Original Post

Hi Steve,

excellent observation and your conclusions are entirely correct. Especially when you want to compare (speech) signal level to background noise, you can't just use the figure from the SPL Tab.

This is because the (relative) spectrum of the speech is applied 'under the hood' for the STI (incl. SNR) calculations. The only way to access actual numbers for the spectrum within Modeler is to use listeners, predict the STI for them and export listener-specific data by right-clicking the MTF graph.

Another way (to get an averaged figure) would be to export the mean SPL from the System SPL tab and post-process that data, e.g. in Excel, which could then also apply any frequency weighting. For this step, you need to know the relative speech spectra for the various STI revisions.

Have you found the STI Appnote on this forum ? It contains a lot of valuable information regarding this matter, although it doesn't state numerical data. The appnote has been announced in this thread. We are currently in the process of authoring an updated version that also includes the latest (to be published) standard edition from 2019, which we added in 6.11. Then, this post is also touching the same issue.

This is a very important topic and it has a couple of more aspects that I haven't yet touched, like predicting SPL maps for custom signal spectra. For now, please digest the links and let's continue the discussion.

best rgds,


Add Reply

Likes (0)
Having trouble signing in?

We recently updated our sign-in procedure and if you have old sign-in data cached, this can create a problem. Please:

  1. Clear your browser cache and cookies
  2. Then close the browser (not just the window)
  3. Open the browser and try again
Thank you

Please make sure that your profile is up to date