extracting dubbed voice from a audio

[Tri P]

[Removed original post]

[jman98]

Short version - basically it CANNOT be done.

Long version - If a very unlikely set of circumstances is in play, it can be done, but people like you who ask this question always interpret that to mean "Yes, with 100% absolute certainty you can do it" instead of "The odds are like 1 in a million you can do it" which is how it really is.

If the voices are on all on the same channel, it CANNOT be done. Period. Doesn't matter what you want to get rid of and what you want to keep.

In the highly unusual one in a million situation where the original audio and the overdubbed audio are on DIFFERENT channels, it's relatively trivial to just keep the channel you want. Since your dubbed track and the original aren't even in the same format, I'm going to tell you that it can't be done. If you convert the AC3 track to MP3 in the exact way that it matches the original audio on the MP3 track, with a good audio editor it might be possible to have them cancel each other out and leave just the dubbed voices, but I don't have the patience to walk you through it since it probably won't work. Sorry.

[newpball]

Try with this

http://www.sonycreativesoftware.com/spectralayerspro

[Selur]

Melodyne might also be a good (not that easy to use) tool to extract the vocals.

[jagabo]

In principle you have the original audio (A) and that audio plus the voice over (A+B). So you can subtract the former from the latter to get just the voice over. (A + B) - A = B. You will lose all music and sound effect too -- but you can make your own mix from the original audio and the cleaned voice over track.

In practice, both tracks have been encoded with lossy codecs, the tracks have different volumes, the volume of the original audio in the latter track varies and has been low pass filtered, and there is a phase difference between the two tracks. So the job will require a lot of work adjusting the volumes, phase, etc.

[Tri P]

[Removed original post]

[jman98]

What's the best software for doing it? Could it be done with audacity?

I see that you have ignored my warning that you're almost certainly not going to be able to do this and interpreted the theoretical "it can be done if you know what you're doing AND you make a million changes AND everything possible goes your way" replies to mean that with 100% absolute certainty you can do this, so I wish you the best.

[Tri P]

[Removed original post]