Stereo Tool

I have Googled a bit, according to this site the normal 19/38 kHz is used increasingly in the OIRT Band - are you sure that this 31.25 kHz is really needed?
http://ukradio.info/OIRT_Tuner/

in September 2015 I monitored stations in Belarus in the low FM band. All were either polar stereo or mono. After talking to a collegue why Mono he told me due to the lack of available transmission hardware. 19 kHz is mostly useless because the radios (except for some Tecsun 64-108 Tuners) decode stereo only in Polar Mode in the lower FM part.

As for the 14dB reduction: This is only for the pilot of 31,25 kHz to save deviation. The rest of the modulated Stereosignal stays at it is.

Now things are starting to make sense. Usually you generate a stereo signal around 38 kHz with no (remains of a) carrier at all. However in this 31.25 kHz modulation you do leave a bit of carrier there, so you suppress it, but not completely. And that formula might be how to do that.

Unfortunately, I still have to implement it to make the phase correct.

Yup. If you fill in 0 for f (frequency), then the formula results in 1/5 which is 13.98 dB reduction, which matches the 14 dB mentioned. So indeed this formula somehow describes a filter.

I think it desribes the Notchfilter of the Carrier and affects also the audio.
This is the curve from 0 to 15 kHz. I hope I can do my tests next days, was very busy on the weekend. We should see it by doing a sweep that when the carrier is not reduced to that formula, that the audio isn't linear anymore but over attenuated in the bass. A bit like RIAA on vinyl....

In the Receiver the Carrier is amplified again by 14dB, so it will be linear again.

Yes. So please try that, and also if it sounds "kinda" ok. If so I can try to add this function. When I figure out how to interpret it (those shapes look a lot like what I had expected, where did you find them?)

I made an Excelsheet and put values from 0 to 15 in 0,1 kHz Steps into the formula for "f". I don't know what j means, I just ignored it

Btw according to the text, f is in kHz, so the graph would end around 15 or 16! Ugh...

I wanted to continue with this but now I really want to wait for your test results again

@radio700, any progress on this?