Since the decoders are different it would be the best to speed-match them (same way you speed-match any MU lashups). When the decoders are both in STOP state then the sound decoder is simply doing its thing (like starting the diesel motor, etc.). In my experience, if there are any delays for the loco to start moving (with sound decoders), those are simply set as a non-zero acceleration setting (CV3), and if there is a delay in stopping then it is a non-zero deceleration setting (CV4). If you set those CV on the non-sound decoder to match the sound decoder that should work. But if the sound decoder in question adds some other type of a delay before the loco starts moving or stops then it seems that perfectly matching it to a non-sound decoder will pretty much be impossible.
