Sorry but I don't know about audiotools. My suggestion is to use REW (longest sweep and higher bitrate) and of course in the receiving room you should get a warning message but it doesn't mean you don't get enough "effective SNR". You check that after in the filtered IR window. If you need more SNR you can use longer sweeps with Voxengo Deconvolver for example. The right way to do it is with spatial averaging and depending on the parameter you want to calculate you may need the receiving room's RT. I don't use REW too often but it should work. I used this technique with Norsonic analyzers (121 and 140) and 100 dB decays wasn't a problem, the head of the ISO working group for 18233 works in Norsonic.
Dodecahedrons are omnidirectional as daisies are round. The problem measuring insulation is not the decay rate but the direct radiation from the source. Standards are intended for diffuse fields and there are always sources of uncertainty.
To use the cross power spectrum typically you need a 2 channel analyzer and a pair of mics placed together. For Microsoft anechoic I think they used B&K 4955 (5 dBA noise). If you don't have the analyzer you can record the signals and use Matlab. The noise of a microphone is V shaped, in one end thermal noise prevails and electrical in the other.
https://aip.scitation.org/doi/pdf/10.1063/1.1149785
Thanks. I'm going to test this. With REW, that warning means it doesn't record anything. However, there is another option, You can export the sine sweep as a wave file and then record it. In fact, I could run mics between the two rooms and record the sweep at the same time at the two mics through a basic 2 channel recorder. Those could be loaded into REW in one of two ways. First, you could convolve them with the original sweep, which gives the transfer function for each room separately. You could then divide the two and that gives you a difference response (the transfer function of the wall and the transmission loss). You could also divide the receiver room wav with the source room wav instead of the original sweep file and that should give you the transfer function of the loss through the wall. I want to try this. I happen to have two identical Dayton measurement mics, so I should be able to do this.
My lab grade mic is a single, I don't have a pair, and I assume you couldn't really do this with mismatched mics. I thought about measuring each room separately, and that might work fine too. I was a little worried about small differences in the capturing of the sweep that could corrupt the response, but if we followed my later comment and averaged multiple sweeps, those small errors would be negligible I imagine.
I've looked for implementations of this, but it seems like almost none of the commercially available softwares have implemented this protocol. They all use the old one I was familiar with. Norsonic and B&K are among the ones I found that do implement it, and a few others out of Denmark that developed the protocols for B&K. The company I bought my mic from sells and rents Norsonic, so I may ask them.
As for the cross power and MS, I looked into it more myself. They used B&K's software, which I don't have. The mic they used has a self noise of -5dB broadband and -15dB in the audio band. I emailed my contact at B&K, but from what I can tell from the literature, this method improves the signal to noise ratio dramatically, but it does not eliminate the self-noise. That if you need quiet, you still need a very quiet mic to begin with. I did see that it has been used (using B&K mics) in a domestic home theater setting, but that person had B&K do the measurements on site. I don't have those tools. I am going to look into Matlab and two mics, but I need to better understand how much improvement this achieves. One study suggested just 10-15 dB's. If that is so, the self noise of the Dayton mic is 25-30dB. My current mic is quieter than that, so I would really want to be sure this is worth the effort.
dodecahedron speakers are common, but I don't have one. In my area around Chicago, I know of no acousticians that work with domestic sound testing that own them either. I know only of large firms that use them. They typically cost in excess of 5 grand. All of my comments were based on the assumption that you too did not have a dodecahedron speaker. It isn't common at this level of work.
Some of the work I found on this approach for TL suggested an omni-source is not needed and instead used standard directional speakers located in the far corner of the room. While this isn't necessarily what the ISO standard states, they seemed confident in the results. It seems reasonable and more doable for something like a home theater or even a small low budget studio/mastering room.
Your comment about the spatial averaging, I saw reference to this being a problem with the method. The studies suggested it wasn't possible, though some authors noted what I would have thought, that you can just take multiple sweeps at different locations and average them. A vector average may be possible, which retains all the information in the impulse in the average.