any way to extract music from spoken word recording

I was recently at a conference where the AE forgot to change his music feed to records from pre-fader to post-fader, had the music playing in the entire time, and recorded the music over the presentation spoken word audio for several days.

Anything that can be done about this?  iZotope RX plugin maybe?  

If the music that was recorded was matched with the music files in a DAW and the phase was flipped, would that work?  I'm assuming it would have to be at an identical location and the volume would have to be matched.  And what a tedious pain in the ass that would be.  But is it possible.  We're talking about a multi-million dollar client here.  

Would a professional audio cleanup house be able to do this?

Thanks!