I would agree with Altec in that it is probably a filtered pink noise being used for calibration. Reason for this is that you want to get as close to a flat response as possible from your sweep. In so doing the deviations from the measured response in flatness are readily apparent. The easiest way to to obtain a near flat result is to utilize a signal with a downward slope of 3db/octave. That of course means a filtered pink noise.