I'd guess that this happens because the clock that drives your output device (RME), is NOT the same as the clock that drives your input device (UMIK).
In such case, there is no relation between the in-versus-out timings, and thus nothing reliable can be said about the phase or time behavior of your measured signal.
(That is : if the software has no tricky workaround for that with timing cues inside the signal, like REW can do. I do not use Dirac, but would be surprised if it used such cues). Not sure about Dirac, but this is a very well-known limitation for speaker builders using gated measurements.
If the above is indeed the cause of your error, then you should be able to finish a full measurement by using a normal XLR microphone (ECM8000 is cheap), connected to your RME.
Both in- and outputs are then using the same RME device, and the same clock.
Also : It would then not be possible for Dirac, to fully solve that (apparent) bug. It can only be worked around in a nasty and not-too-reliable way, by hiding timing 'blips' inside the measurent sweep, and making a few assumptions that may or may not always be true. Or by simply ignoring time related stuff, which previous versions of Dirac <=v3.4.4 seem to do (at least, that's my assumption when I see the suggestion on net to install v3.4.4 as a 'fix' for this problem...which, users say, create the exact same measuments but now no error message is thrown.).
Again : not familiar with both dirac and this error message, so I can't be sure. But it seems very logical to me....
OH : I see you tried using your laptop's on-board mic as well, and that it worked correctly. My assumption is that you were using your laptop's on-board sound OUTput in that case as well. Not the RME out. Is that correct ?
In such case, there is no relation between the in-versus-out timings, and thus nothing reliable can be said about the phase or time behavior of your measured signal.
(That is : if the software has no tricky workaround for that with timing cues inside the signal, like REW can do. I do not use Dirac, but would be surprised if it used such cues). Not sure about Dirac, but this is a very well-known limitation for speaker builders using gated measurements.
If the above is indeed the cause of your error, then you should be able to finish a full measurement by using a normal XLR microphone (ECM8000 is cheap), connected to your RME.
Both in- and outputs are then using the same RME device, and the same clock.
Also : It would then not be possible for Dirac, to fully solve that (apparent) bug. It can only be worked around in a nasty and not-too-reliable way, by hiding timing 'blips' inside the measurent sweep, and making a few assumptions that may or may not always be true. Or by simply ignoring time related stuff, which previous versions of Dirac <=v3.4.4 seem to do (at least, that's my assumption when I see the suggestion on net to install v3.4.4 as a 'fix' for this problem...which, users say, create the exact same measuments but now no error message is thrown.).
Again : not familiar with both dirac and this error message, so I can't be sure. But it seems very logical to me....
OH : I see you tried using your laptop's on-board mic as well, and that it worked correctly. My assumption is that you were using your laptop's on-board sound OUTput in that case as well. Not the RME out. Is that correct ?