Reduce the amount of kinetic energy loss in a collision

[fdajkffk]

I have a lab in which a moving marble collides with a still one. What changes could I make to reduce the amount of kinetic energy loss in the collision? The track for the marbles is a piece of card stock paper.


I was thinking:
-Use two tubes as a track which will reduce the amount of friction on the marble and prevent a glancing collision.

How would I eliminate energy loss to sound and heat? Any ideas?
Any other ideas for reducing ek loss as well?

 

[haruspex]

I assume that the objective is to transfer as much energy as possible to the second marble, rather than to minimise the energy lost. I.e., you don't care about energy retained by the first marble. So yes, ensuring the collision is head-on is good.
Would it be possible to put something between the marbles? E.g. a thin piece of rubber, maybe from a superball? The downside of that is the high friction, so you might need another (hard) layer either side of the rubber with low friction.
Another possibility is to put a third marble in between, touching the stationary one. The theory behind this is that when a rolling ball strikes a stationary one it tends to impart backspin. The extra marble in the middle will reverse the spin.