Hello! I work for a sound masking company and they've asked me to play around with REW to test their sound masking systems. I was running a frequency response test on one of their white noise contours and noticed that changing the length of the recording changed the SPL that would be recorded on the graph. I've attached an image below that shows this. The green curve has a length of 128k, the purple 512k, and the blue 4M. Please let me know if anyone has any information on why this would be occurring. All of the curves were taken with the same speaker output.