When we look to achieve ultrasonic muzzle speeds, we will need propellants that can burn at a rate such that they can fill the void created by the traveling projectile much faster than it creates it, otherwise they will not continue to apply pressure to the projectile as it travels down the barrel.
Actually, that isn't true. You could use slower-burning propellants and simply create a very large cartridge that is necked down to the bore diameter. More slow-burning propellants gives off the same amount of gas and less faster-burning propellants.
So, we've established that the burn rate of conventional propellants isn't a limiting factor because we can simply increase the amount of propellant that is being burned at any given time. We are temporarily ignoring the effects of the local speed of sound in the chamber and barrel. This, then, leaves us with the strength of the material as the only thing that limits muzzle speed, because it limits the pressure that can be contained.
"Parts which are expendable may be designed to deform but not rupture at transient pressures as high as 1,000,000 PSI," according to the paper to which DYI linked us. (Page 19)
So, we would have to design a system with a satisfactorily high maximum pressure (anywhere from 100,000 PSI to 1,000,000 PSI) and arrange an amount of propellant such that immediately upon ignition, the chamber pressure reaches this maximum pressure but does not exceed it. To maintain this pressure, we would have to ignite additional chambers such that an amount equal to the original amount of propellant is ignited for every volume of the barrel equivalent to the chamber volume that is traversed by the projectile. This will essentially maintain an average of 75% of the maximum allowable pressure for the entire barrel.
Anyone with a machine shop and a near-by RadioShack could create this system, which would be capable of unearthly muzzle speeds with the proper selection of bore area, barrel length and projectile weight. Again, I'm completely ignoring the effects of the local speed of sound, which I think would ultimately be the limiting factor.
So, to narrow it down again, what advantage could an ETG have over the previously mentioned system which could use any conventional propellant?
Here is some truly next generation stuff:
http://arxiv.org/ftp/arxiv/papers/0910/0910.3961.pdf