To start with, you will not be able to do this completely. There will be TV sound, and most likely the voice will sound weird. I will be guiding you in methods to manually do this, but you can find programs that will also partially help.
So the first thing would be to get a compressor and expander (audacity has both, and it's free). First use the expander by increasing the dampening, or attenuation until you hear it actually reducing volume. Then fiddle with the threshold (where the attenuation kicks in) so that it doesn't reduce the volume of the subjects voice. Then you can do sort of the reverse of that to increase ONLY the subjects voice.
(At this point, if I've lost you, then you should go with one of the programs, or look find an audio engineer to do this for you)
Another part is filtering out the TV as much as possible with EQ. First crank the amplitude up all the way, and narrow the Q value (the bandwidth of frequencies effected). Then you can sweep this up and down in frequency. Find where the TV sticks out the most, and then take the amplitude all the way down and widen the Q just a little bit. You might need to find multiple points to do this. You can then do similarly in trying to raise frequencies that strengthen the subjects voice.
A High-pass filter might also help, depending on a lot of things.
This may or may not help you, but I hope you get what you're looking for.