Detection of arrival time does not seem to be right, why?

On of our customers reported:

“Here you have two measured impulse responses measured using the MD player. The arrival times are not corrected as you can see. Around the impulse you see some artefacts, probably due to the MD player. When I play the original wave file over another sound card the impulse response is cleaner, but this is good enough for acoustic measurements.”

In the lower of the two figures above, we see the impulse response plotted on a dB vertical axis. When doing this we see that the impulse response has a very long pre-ringing. This is typical and is probably due to the artifacts of the MD player

In the frequency response plot, we also see that there is a dip at 7 kHz in the red curve. Such a dip should be avoided if using the method of a single sweep and starting the measurement just before the sweep starts.