Musician Pegasus DAC claims 32bit/1536kHz with STM32F446 USB receiver. There are several projects on diyaudio for STM32 UAC2 device firmware, perhaps their authors (e.g. @bohrok2610 , @slerpxcq https://github.com/slerpxcq/openuac2_fw ) would tell you more.
1536k means the I2S data line will run at 100MHz.
Also UAC2 will need multiple transactions/high bandwidth per USB2 microframe - 1536 * 4 * 2 / 8 = 1536 bytes per microframe which requires two isochronous transactions (1024 bytes max) per microframe. E.g. iOS devices do not support this https://forum.rme-audio.de/viewtopic.php?pid=178246#p178246
1536k means the I2S data line will run at 100MHz.
Also UAC2 will need multiple transactions/high bandwidth per USB2 microframe - 1536 * 4 * 2 / 8 = 1536 bytes per microframe which requires two isochronous transactions (1024 bytes max) per microframe. E.g. iOS devices do not support this https://forum.rme-audio.de/viewtopic.php?pid=178246#p178246
One may ask, what is the use of 32bit/1536kHz? This is just a waste of data and effort. We are in the audio domain.
A practical useful application would be measurements and analysis. In audio - marketing (which is perfectly fine, IMO).
It is quite possible to run STM32H7 UAC2 at 32bit/1536. The challenge is that I2S timings will be very tight and would require careful tuning at USB-I2S board. E.g. my boards have fast I2S isolators but even those may struggle at that rate. And not many DACs or ADCs support 1536k.
The AK4499EXEQ supports 1536k, I would like to find a method to actually test this. I have already been able to confirm up to 768k.
Hopefully the newer AKM DACs implement the 384kHz+ rates better than AK4493 https://www.diyaudio.com/community/...ling-speeds-oversampling-filter-issue.372469/