Well, that's a different way of approaching it. If you look at the reactor pictures in the third link, you'll notice that the co2 is injected after the output of the pump. The output current is supposed to twirl the co2 bubbles while their natural tendency to float up and the sponge keeps them in that section until they get fully mixed into the water before it gets out of the bottom. There'll be some minimal amount of bubbles escaping, sure, but the bulk of them should get mixed in. The venturi loop is only in case there's too much buildup, and then some goes back to the input impeller of the pump. So, yes, there's probably noise at that point, but not all the time.