Ronster said:

brash_stryker said:

I've been playing around with isolating dialogue from music using the spectral repair setting in Izotope RX 4, but it's really difficult. Seems to be trial and error for me because the visual representation of the waveform doesn't give you a clear distinction between sounds.

There is no distinction between sounds in a mix

Please know what you're talking about before making authoritative statements like that. Just because sound is mixed together doesn't mean there's no distinction. If there weren't a distinction (in pitch, frequency etc), then software to try and isolate one sound from another wouldn't exist.

For the simplest example possible, look at the pattern the police siren makes in this video:

https://www.youtube.com/watch?v=xQucdzpL0tc