What you want is called a compressor. It will work as long as both voices aren't speaking simultaneously. Basically what a compressor does is that you set a threshold level and then once anything passes that threshold level, you set a ratio of how much louder you want things to get. So for example, if you set a 2:1 ratio, then everything louder than the threshold would only get half as much louder as it otherwise would.
If you set to ratio to infinity, it becomes a limiter as nothing can exceed the threshold. I'd suggest trying to just use a high ratio of compression though (like 6 or 8 to 1) and see how that goes. Hard limiting can have undesirable impacts on the way the voice actually sounds since there are natural dynamics in our speech itself.