Others with more experience of room correction may want to chip in, but there are several approaches of increasing complexity.
Assume that you are using stereo, then:
You can measure the impulse response of your left and right speakers separately, and apply the results to the respective channels. The first example on http://convolver.sourceforge.net/configegs.html
shows the convolver config file that allows you to do that. Alternatively, you can combine the mono impulse responses into a single stereo WAV file and load that directly instead of using a config file.
That is not, however, the theoretically best way of correcting your system, although it may give good results in practice and has the merit of simplicity. (At least not if you don't listen with a board in front of you that prevents the left channel sound from reaching your right ear ...)
The more correct way, apparently, is to take 4 impulse response measurements (left speaker to left ear, left to right, right to right and right to left) and then apply them using the head-related transfer function (HRTF) network example on http://convolver.sourceforge.net/configegs.html
as a model.
I'll leave it to others to comment on / correct this advice, but the long and the short of it is that with convolver you can mix input channels, filter the result, mix/sum the results of different filter paths, and direct them to specified output channels in a completely configurable way. (In fact, I am trying to decide how best to reduce the flexibility, as it makes convolver more complex to set up than is probably strictly necessary.) So, if you want to do other things like split a channel into different frequency ranges and send the results to different output channels, you can do it.