Get background music from two different language tracks

[Blue Mantis]

I've been absolutely racking my brains over doing this for hours now and got various VST plugins that don't seem to help. Theoretically, there is absolutely no reason why this shouldn't be possible. The music is identical but the voices are different. One is in English, the other Japanese. Just the voices differ, and the music is at the exact same position. Heck, I even have the Spanish DVDs. So if it would help, I could use a third language to filter out the differences.

If somebody can please help me with this, I'd seriously appreciate it. You can find the two audio files here: https://www.dropbox.com/sh/hua2bsn8n...zpHwpnAVa?dl=0
If you can crack this problem, please tell me how you did it.

Thank you!

[cwsand]

Originally Posted by Blue Mantis

I've been absolutely racking my brains over doing this for hours now and got various VST plugins that don't seem to help. Theoretically, there is absolutely no reason why this shouldn't be possible. The music is identical but the voices are different. One is in English, the other Japanese. Just the voices differ, and the music is at the exact same position. Heck, I even have the Spanish DVDs. So if it would help, I could use a third language to filter out the differences.

If somebody can please help me with this, I'd seriously appreciate it. You can find the two audio files here: https://www.dropbox.com/sh/hua2bsn8n...zpHwpnAVa?dl=0
If you can crack this problem, please tell me how you did it.

Thank you!

Hi, You're trying to extract just the music? That's going to be very difficult if not impossible. You can try to minimize the dialogue with processing, and maybe edit between the different versions to have the least amount of dialogue, but you can't completely eliminate a sound overlayed on a single file.

[MatzeHD]

This is a big problem, you can only make this:

If you make up the difference signal from the stereotrack, you have after this only the S-signal.

L+R=M(ono)signal
L-R=S(ide) signal

I've produced this with your wav files, it works, but with this methode it lacks all other monaural signals too, like bass etc!

A other problem is, your sourcemix is not so good, its lossy audio, you mean its was from a dvd and the mix is not clean, channel seperation is bad and some sounds from the M-signal are available too in the S-signal!

Ok, i hope you understand my message, its difficult to explain for me in my bad english!

[amagras]

I've heard good things of Izotope Rx... I've no idea if it's been designed to deal with this kind of situation but you can give it a try.

[Ben Jenssen]

Originally Posted by Blue Mantis

Just the voices differ, and the music is at the exact same position.

Hi.
I thought I'd give it a shot so I set up a session with the two files.

Thought I'd flip the phase on, say the english version and mix it with the japanese so that the music would cancel out and I'd have a track that contained the speach only, and this could later be phase flipped, mixed in with original tracks to cancel out the speach.

But, the waveforms, where they contain only music, are not identical. It's obvious when you zoom in. So, not possible I'm afraid. I couldn't even get close.

[sw rec]

Think of mixing yellow and blue paint together to make green. Once you do, you can't take the blue paint back out. I don't know of a way to "unmix" anything, not convincingly.
The music is identical but the voices are different. One is in English, the other Japanese. Just the voices differ, and the music is at the exact same position
Try editing instead, maybe?