justme, I wish I could do more testing, but if I posted screen shots of the two frequency graphs from the recordings, you'd instantly see the difference. Definitely not in the faux audiophile expert realm that is, 'the sound stage is more balanced and lean yet crystaline in clarity and presence' here, it's: one sounds like the recorded source, and one sounds like crap. With the graphs showing the corruptions clearly.
Ozpeter, one thing that USB Audio Recorder Pro does is they made their own usb audio driver, probably based off the Linux kernel's normal driver is my guess, they don't use the usb audio driver of Android. This is actually quite meaningful, because if that was not the case, then it could be different Android versions shipping different or varying usb audio drivers, which would certainly result in such issues. This usb driver and logic is used on their Evolution DAW as well, so it should be working for a while, I hope.
But you are quite right that this is unfortunately not 100% knowable since Android is something of a blackbox sadly, except for the fact I have done many fine recordings with the other usb device on the same phone at high db levels without a single issue like this, when I have gain too high, the recording clips, period. I cannot make the Rode clip. I've had issues at high db levels, but they are what you'd expect from either overloaded mics or overloaded circuits in battery box or elsewhere.
One reason I don't actually think it's even worth testing the other usb recording apps for android is they are all worse, worse interfaces, worse usabilty, worse documentation, some even try to force you sign up to their 'cloud' stuff (BandLab) so I couldn't even test those at all, but I assume they are as bad or worse. The dolby one was particularly terrible, which was a bit of a surprise, that one doesn't even pretend to not do dynamic processing, in fact they brag about it, and it's clearly not something that can be disabled.
I could test another round recording straight to a real computer, same tests, but as noted, my speakers just don't like going to just below small club levels, and blowing their woofers isn't worth it to me since I already see the non ambiguous data in the recordings I tested with.
Everything points to Rode making very bad decisions about their firmware, and also to a certain NIH (not invented here) attitude towards purchasers of their AI-Micro, since none of those filters or settings should be active in the first place when not going into a Rode app.
The clear presence of a bass cut as well means I can't even trust their own docs that say clearly that if you disable high pass filter in their app, it will disable it on the device for 3rd party apps as well, which does not appear to be the case. Or they messed up their firmware, and it's supposed to work, but doesn't, it is too hard to tell to be honest, and not one of these issues should exist on what is supposed to be a quality device.
I'm keeping my fingers crossed for a few things now:
* The old usb audio interface performs as expected (and given it's coming in around 10db or so under the Rode, which is roughly the range that was being redlined with the rode, I have good optimism there). I'll know this tomorrow.
* A serious usb DAC maker, like Dragonfly, decides it's worth making a quality stereo mic input ADC. This is unfortunately a very fringe requirement, so the market is clearly not very big, but I'm going to keep my eye out, as we see with Rode, the chips are there, but all that's needed is either for them to wake up and disable all their processing and deliver clean audio via a firmware update, or for another vendor to make a stereo 24/48 capable ADC using the same chips, unless the processing is built into the chips, which I suspect is not the case. There is nothing in the tech requirements at all that makes this not possible, so it's just a matter of someone doing it, and deciding there's enough market to make it worth it.
After this experience, I think I'd definitely bump up my maximum price I'd pay for such a device however.