I put together my own cover of a musical composition from How to Train Your Dragon, which I intend to put on YouTube. I want to sync it up with the scene from the movie where it plays, but I thought it would interesting if instead of just replacing the audio, I could put the dialogue from that scene on top of my music.

Of course to do that, I would have to isolate the dialogue audio from the rest of the movie soundtrack, but how hard could that be, right?

I've included links to two audio clips which I've cut directly from the movie. They haven't been re-recorded or anything, so they should be the exact waveforms from the film, cut from exactly the same time. One of them has English dialogue, the other has Spanish.

English Clip
Spanish Clip

My theory was that the two clips should be exactly the same except for where the dialogue occurs. I thought I should be able to:
1) Take the left audio tracks from both stereo clips.
2) Pan one hard left and one hard right.
3) Use mid-side processing to separate out what's the same between the two tracks (the music and sound effects, which would be the mid channel), from what's different (the overlapping Spanish/English dialogue, which would be the side channel).
4) Subtract the mid channel from the original English clip and be left with only the English dialogue. 5) Repeat for the right stereo tracks and then put them back together.

This approached failed at step 3. Looking at the two waveforms (from the two languages) side by side, you can see that they are actually different, even in the parts where there is no dialogue, despite the fact that those parts sound exactly the same (at least from what I can hear).

This is where I'm hoping you can help. Can you tell what's actually different between these two files? Or even provide an educated guess? If I can figure that out I might be able to account for it.
Otherwise if you have another suggestion on how I can isolate the dialogue, I'll take it. But before you suggest it, I already tried going through the 5.1 surround-sound channels and none of them have the dialogue by itself.

  • Grab the 5.1 Blue-Ray, rip the audio and grab the centre channel? The Surrounds won't have what you are looking for, but the centre channel should have all the dialogue you need.
    – Mark
    Feb 18, 2019 at 12:45
  • @Mark As I said at the end of the question, I already got a hold of the 6 channels from the 5.1. The center track has music mixed in with the dialogue. Feb 18, 2019 at 15:08
  • That's unfortunate. Sorry I interpreted your statement as having inspected LsRs only.
    – Mark
    Feb 18, 2019 at 22:22

1 Answer 1


Audacity has a plugin to remove or isolate voice - Vocal Reduction and Isolation which I haven't tested. Izotope RX's Music Rebalance does a fair job of it, but that's expensive.

My guess is that both your audio sources come from differently compressed soundtracks & therefore won't match exactly.

  • I hadn't thought of compression, that's probably what the difference is. The Audacity plugin basically does what I was trying to do, but with a single stereo track: isolate the center vocals/dialogue with mid-side processing, except it also adds a LPF and HPF to help narrow things down. It doesn't really work for this one. Izotope is a bit to pricey for a fun side project, especially since I don't know if it will work. Feb 18, 2019 at 15:30
  • I did a quick try on the English version - Izotope 'works' for a loose definition of 'works'. You'd get away with laying it back with a solid music track. In isolation it's nothing to write home about, though it's a whole lot better than could be done even just a few years ago.
    – Tetsujin
    Feb 18, 2019 at 15:32
  • Well, without Izotope I guess Audacity would be the closest I could get. Feb 18, 2019 at 18:17

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.