logo Sign In

Help Wanted: The Prequel Edit Problem - remove the music & sound effects without removing the dialogue... — Page 3

Author
Time

Just posting here, to say I'll post my attempt at music removal sometime later today or tommorow.  I do think it is a little more possible then it is thought to be.

On another note, that spectral sound editing I mentioned earlier is being taken to new levels in this sneak peak for a future Adobe version a year or so away.  While they are basic files he's playing with, the fact that there are no artifacts is amazing, and shows we are getting closer.

http://tv.adobe.com/watch/adobe-technology-sneaks-2013/sneak-peek-audio-layers/

Author
Time

emanswfan said:

On another note, that spectral sound editing I mentioned earlier is being taken to new levels in this sneak peak for a future Adobe version a year or so away.  While they are basic files he's playing with, the fact that there are no artifacts is amazing, and shows we are getting closer.

http://tv.adobe.com/watch/adobe-technology-sneaks-2013/sneak-peek-audio-layers/

That's incredible! This could be a total game changer for SW edits.

Author
Time

It seems that ROTS opening is an extreme chalange to try to get rid of the music, and after tweaking around through multiple different software, I only got a result about the same (but without the dialogue gating issues of that video) as previous posted video.  However many non-action places are easily isolated.

I did goof for hours around enough to discover: if one wants to remove the music from any of the action aequences, they'll need to use a spectral display to extract each peice of dialogue or necessary sfx and combine it with new sfx/foley along with the new music.  The dialogue and some sfx can be perfectly extracted from the music but only through manual spectral masking, which I don't feel like doing.  I have a limit to my strong work ethic towards all this, I'm not like Ady and his ridiculous rotoscoping.

Knowing this though, I don't doubt the fact that feature I highlighted above would work with Star Wars, so if a year or so wait is what is necessary so be it.  It clearly just automates most of the spectral extraction process.

Author
Time
 (Edited)

It would take a couple of hours per 10 minutes of film, but the vocals could be cleanly chopped from the soundtrack, moved around in the spectrum until the music is as minimized as possible, and then judiciously manually re-pitched to match syllable by syllable.  Same with the sound effects.  It would be an incredible amount of work, but I do this kind of thing regularly to remix songs.  I don't mean to sound like a know it all, (especially on what is probably my first or second post here ever,) but I just don't know of a shortcut that would ever sound as good as the elbow grease needed to fully isolate these things from one another.   

Author
Time

Proven again, rescoring helps the movies so much.

Also, BTW I have been trying to get back at this with several new methods of music/SFX removal that I've come up with myself.  They're really experimental, but if I can figure out how to execute them just right, I may be able to isolate the vocals perfectly.

I'm repairing my computer this week, so I'll get to work on it soon, and hopefully I'll get much better results.

Author
Time

MrInsaneA said:

http://www.youtube.com/watch?v=6PThahkUwY8&feature=youtu.be 

 

I'll just leave this here.

Wonderful! Music fits like a glove.

btw Would love to hear an editor remove the bizarre race-car overdub at 1.09. I think Ben Burtt was drunk on his own power and thought he could get away with anything on the prequels. He must have caught something from GL.

VIZ TOP TIPS! - PARENTS. Impress your children by showing them a floppy disk and telling them it’s a 3D model of a save icon.

Author
Time

Yeah, the music is perfect! Same for the Attack of the Clones clip. 
It's amazing just how much atmosphere you can add just by changing the music. 

And I don't think Ben Burtt was ever drunk on any power. The only issue is that he didn't do nearly as much original work for RotS, using almost entirely existing sounds, so some of the sound design is somewhat bland or uninspired.

Author
Time

Just an amendment, but MrInsaneA, I love your solution to getting Grievous out of there without the whole window exploding thing!

But I felt it was cut kind of awkwardly. 
Maybe this might work in a more fluid way?
http://vimeo.com/70733599

Author
Time

aalenfae said:

Just an amendment, but MrInsaneA, I love your solution to getting Grievous out of there without the whole window exploding thing!

But I felt it was cut kind of awkwardly. 
Maybe this might work in a more fluid way?
http://vimeo.com/70733599

Wow, my eyes just melted from the awesome genius of that edit. You gotta use that one!

VIZ TOP TIPS! - PARENTS. Impress your children by showing them a floppy disk and telling them it’s a 3D model of a save icon.

Author
Time

Aalanfae, you're a genius, and a scholar. 

Author
Time

Haha, I'm simply expanding on your idea!
I've totally cut this scene from my particular edit, but seeing how you've improved it with clever cuts and music... I almost want to put it back in!

Author
Time

Signed up just to post this.  It's not the most elegant way but if you can find player (br or DVD) with multi channel out, you can have more success.  I know i had a br player with discrete outputs so I could get uncompressed audio in my home theater.  Aldo, i have a background in audio production and have a multi channel audio input board in my production setup.  It would be time consuming to say the least but if a multi-channel out were used, each of the five channels could be captured real-time and recorded on their own separate tracks.  The dialog channel is more often than not just dialog.  Hope this helps and if it's been covered already just forgive me for being too lazy to read all three pages.  I have a session coming up. Sorry!

 

cheers

chrid

To contribute to The Starchaser 3D project, please pm for details

Author
Time

I don't understand what good multi-channel out on a player would do.  Everyone can already separate the source audio into multi-channel.  The problem is that with Star Wars, the "dialog channel" (which is actually the center channel) has lots of music and sfx in it as well.

Author
Time
 (Edited)

Okay, this is a little theory of mine,

I did not try it yet, and since I'm not a math guy at all, I could very possibly went wrong with my reasoning.

So...

IF, and only IF center chanel (audio english) and (audio french) share the same EXACT mix (and it is rarely, if never, the case... so I guess all this theory means nothing in the end, lol) then:

Step 1: (audio english) + inverted audio french) = English voices mixed with inverted french voices

Step 2: (audio english) + (English voices mixed with inverted french voices) = (audio french (with french voices being inverted)) mixed with (english voices at volume level X2)

Step 3: (audio french (with french voices being inverted)) mixed with (english at volume level X2) + (audio french) = (audio english (with voices at volume level X2)

Step 4: (audio english (with voices at volume level X2) + (inverted audio english) = .... English voices at original volume level.

Step 5: (audio english) + inverted (English voices at original volume level) = (audio english) minus english voices...

 

EDIT: All this reasoning is based on the supposition that an inverted audio removes half of the same original audio if its level is twice the same. And this is not a given.

RE EDIT: mmmm step 3 obviously double the music.... I must return to my reasoning, lol.

Author
Time

TMBTM said:

Okay, this is a little theory of mine,

I did not try it yet, and since I'm not a math guy at all, I could very possibly went wrong with my reasoning.

So...

IF, and only IF center chanel (audio english) and (audio french) share the same EXACT mix (and it is rarely, if never, the case... so I guess all this theory means nothing in the end, lol) then:

Step 1: (audio english) + inverted audio french) = English voices mixed with inverted french voices

Step 2: (audio english) + (English voices mixed with inverted french voices) = (audio french (with french voices being inverted)) mixed with (english voices at volume level X2)

Step 3: (audio french (with french voices being inverted)) mixed with (english at volume level X2) + (audio french) = (audio english (with voices at volume level X2)

Step 4: (audio english (with voices at volume level X2) + (inverted audio english) = .... English voices at original volume level.

Step 5: (audio english) + inverted (English voices at original volume level) = (audio english) minus english voices...

 

EDIT: All this reasoning is based on the supposition that an inverted audio removes half of the same original audio if its level is twice the same. And this is not a given.

RE EDIT: mmmm step 3 obviously double the music.... I must return to my reasoning, lol.

You're workflow is getting there, but I'm not sure it's perfectly right.

In regards to the mix, no mix in the prequels are the same.  On the original DVD's, other languages were 2.0 and are now 5.1 on the bluray.  But since the english track is 6.1 DTS-HD (which has more dynamic range), is certainly not the same mix.  The original DVD's used a 5.1 dolby mix, but those are different movies, the runtimes don't add up exactly.

The new 2013 rereleases might have a matching mix, as the DVD's in the combo pack require a 5.1 English mix, and the disc is said to have the other languages in 5.1 as well.  So the new set, might work.  Though, I won't be entirely certain till the full press release, and even then who wants to buy the movies again (though if this method really works, it would be well worth it.)  I do have a non-star wars bluray, where I know the english and french DTS-HD mixes match.  I'm using it to test the theory.

Author
Time

emanswfan said:

(though if this method really works, it would be well worth it.)

Nah, it's not working, step 3 is a mess. It makes the music X2, same as the voices. Does not work. 

Author
Time

Step 2 is interresting though. Having english volume at volume level X2 could be useful to use removing filters while keeping the english voices. Only little problem would be that french audio is now in the mix and adding human voices in the mix is never a good idea. But since french would be at normal volume, english should be louder and so maybe easier to filter than on the original english audio track.

Author
Time

If Disney/Lucasfilm were smart/cool, they'd release Prequel "kits", with everything separated, and every shot, final cut or not. Hell, they could get crazy and release green screen alts as well. The best thing about the prequels, and Episode I in particular, is the fan edits. Part of the problem with the Phantom Menace is inherent, but another part is the imaginations of so many viewers. Fan edits allow people to at least inch it a little closer to what they wanted, and I think the powers that be should encourage this.

 

It's well past time to give up on the idea that TPM is equal to Episode IV, or whatever fiction Lucas, etc have been telling themselves. It's not a total failure, but it actually could be an inspiring success if they would open up the vaults for fans.