If I have:

  1. audio file someone has created by placing a monologue/speech over background music

  2. an audio file of the exact background music which was used above

is there any way to "subtract" the background music (2) out of file (1) to be left with only the monologue/speech?

I found this thread which is similar, but a more difficult problem. It was also asked over 10 years ago, so I hope it's okay that I started a new post.

  • To do a perfect raw subtract, you need to have (originally) sample-identical tracks - any lossy data compression, aliasing, dithering, or any other distortion, and it will need further processing in something like RX, as Tetsujin said. Also, make sure they really are opposite phase and at exactly the same level. I'd love to know if it works, btw. I have a feeling you'll have problems - it's rarely straight forward. Maybe let us know in chat
    – n00dles
    Commented Jan 26, 2023 at 16:20

1 Answer 1


The solutions are pretty much the same as they were a decade ago, but the AI component has improved a lot. If you don't have room ambience to contend with, this should be relatively easier.

If your two music files are identical then you can phase-reverse one against the other. What remains is the non-identical… the voiceover.

The AI component in Izoptope RX [now up to version 10] could be used to try isolate it. It can understand what is 'a voice' so knows what to try remove to isolate it.

Both will require quite a bit of trial & error, I would imagine.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.