Adobe Audition: Seperating background from foreground?

Snake · Sat 07/11/2009 18:39:19

I was wondering how I go about this. I've got quite a few old recordings of me playing guitar (captured by external microphone) and I would like to be able to seperate the audio so I can hear what is going on in the background (such as birds, planes and people talking). Is there a way to go about this in audition? All I can think of is when they do it in CSI, heh. I know how to eliminate the noise from the audio by capturing a noise reduction profile, but from there I'm stumped. Is there a special plugin that does this?

Thanks in advance, I've actually been meaning to ask this for over a year now. I've thought of it now since I'm fiddling around with some old files as I type this.

Nikolas · Sat 07/11/2009 19:18:13

I'm actually not aware of any specific plug-in which does exactly that.

You could try x-noise or something like that (Waves plug-in) and put it to reduce the music instead of the hissing, but it wouldn't work too well.

One other idea you could, potentially try would be to try and eliminate everything panned and leave only the center (or the other way around). The idea is that if you wrote it in stereo (with 2 mics, which is improbable) you could reverse phase, add, and take out everything in the center, leaving only panned sounds. So if you had the guitar in the center, most of it would be gone, but anything hard panned would stay 'intact' with a phase reverse. *

That's all I can think right now.

* My description is probably not accurate, but I don't have any more time to think right now, sorry...

Leon · Sat 07/11/2009 19:35:16

I know that there are voice removing plug-ins. They don't really remove the voice but eliminate certain frequency's and you might be able to play around with it and find your guitar 'waves'.

According to this article there should be something like "the center channel reduction filter" standard in audition. A tool I've found is Extra Boy Pro that you might want to check out.

Oh, and don't believe everything you see on CSI. Although they use advanced technologies doesn't mean it's all real. Have you ever noticed how they blow up an image of a CCTV footage about 500% and still get perfect quality images? Yeah, you can do a lot with enhancements but there are limits...

Snake · Sat 07/11/2009 19:40:53

Bah. Only one mic was used. So I guess if it's not stereo than it's just one audio wave, eh? Just as I thought, I suppose. I wish there was some way to "lift" the sound of the guitar off and only hear the background.

So what is this "x-noise"? Is it the noise reduction? Or is it a specific plug-in already installed with Audition?

Haha, I know it's fake, Leon, but I was hoping there was something similar out there

nihilyst · Sat 07/11/2009 20:51:50

The only way I could imagine is to filter all the guitar's frequencies, but that would certainly mean killing most of the background stuff, too. What do you need it for, anyway? I could understand if you wanted to GET RID of the background to hear just you playing guitar (but that depends on your guitar playing skills, I guess

). But the other way around, what do you need it for?

Perhaps there's another way ...

Snake · Sat 07/11/2009 22:04:23

Quote from: nihilyst on Sat 07/11/2009 20:51:50
The only way I could imagine is to filter all the guitar's frequencies, but that would certainly mean killing most of the background stuff, too. What do you need it for, anyway? I could understand if you wanted to GET RID of the background to hear just you playing guitar (but that depends on your guitar playing skills, I guess ). But the other way around, what do you need it for? Perhaps there's another way ...

I really don't need it for anything, at least that's important to anybody else, anyway. It's just something I thought would be neat if I could hear the background. While at least two of them were being recorded, it was after a family event, something like Thanksgiving or a birthday. I can hear people talking in the other room. Just thought it would be cool if I could hear what they were talking about, and then I could pin-point when this was recorded. The dates have been changed since I've had them saved on various computers/devices, cropping them and what not. Most of the dates are the actual date they were recorded.

Just thought it would be neat to know what was going on in the background, that's all. Nostalgia of playing guitar in my bedroom at my mums house years ago, maybe

TerranRich · Sat 07/11/2009 22:06:05

To be fair, CSI is more science-fiction than drama. Most of the stuff they do, especially the sound editing they do, is mostly magic and special effects. There is no simple way to separate background noise from foreground sound, simply because of the way the sounds are stored. It'd be like having a JPG of a man with 50% transparency overlayed on a background. You can't separate the two. Sure, you can lasso-tool the man out, but you'll still see background behind him because he was only 50% opaque when the JPG was saved.

Nikolas · Sat 07/11/2009 22:09:39

A sidenote: A WAV file CAN be stereo. The music you listen on your CD is stereo and a single WAV file. But there are also mono files. You could also record a single mic onto a stereo WAV file, but it would result in identical left and right, so it's no use.

x-noise is a VST plug-in made by WAVES.

Terran: Not so fast! While I agree CSI is filled with over*** it still remains that now, in music, you can grab a single audio file (For example a dead guy playing a piece in piano), identidying all notes, and altering then at will. The equivelant to art would be to grab a picture, depict all colours that were used and altering them at will. While it's in infantry, it IS possible (By DNA in Melodyne software, by Celemony).

There are also reports of a software distinguishing different timbres, thus analysing orchestrations. IF one can grab the violin out of the whole orchestra, then it would be possible to grab the background voices and not the guitar...

TerranRich · Sat 07/11/2009 22:14:07

Altering pitch on a note-by-note basis sounds do-able, and within the realm of possibility. But taking a grainy security video, blowing up a tiny section 1000X and having it come out in HD quality is just... not.

Nikolas · Sun 08/11/2009 06:09:38

Quote from: TerranRich on Sat 07/11/2009 22:14:07
Altering pitch on a note-by-note basis sounds do-able, and within the realm of possibility. But taking a grainy security video, blowing up a tiny section 1000X and having it come out in HD quality is just... not.

Actually no it isn't! And according to research what I mentioned above was simply considered IMPOSSIBLE.

Think about it: You have a vinyl record of... Rachmaninov (long dead) playing one of his pieces! You take the vinyl, ALTER THE INDIVIDUAL notes, and then you end up with a new piece, but FROM THE PLAYING OF THE DEAD GUY! It's mind blowing and the parallel is pretty much to grab Da Vincis pics, depict which colours he used and alter them, or depict each stroke he gave with his brush and alter those. Think about it!

TerranRich · Sun 08/11/2009 06:52:42

What I'm talking about here is what is possible vs. what is impossible. It is not possible to separate foreground from background in audio, for the reasons I mentioned. Just re-read my comparison to a JPEG with a semi-transparent man in front of a background.

As for the alteration of individual notes, that is definitely possible, because it's all about mathematics. Each note has a certain frequency, and even combinations of notes can take the shape of a certain waveform. It all boils down to mathematics and waveforms. But background noise, which can basically be considered white noise, is impossible to separate out, unless the foreground sound is something absolutely simple, like one instrument playing notes.

The Bedminster Incident · Sun 08/11/2009 10:29:32

Everything comes down to maths. So everything is possible. I'm gonna go build a lab now and create a time machine.
Actually, that might not be too far from the truth, seeing as people considering it impossible to fly.
Let me just make up a theory: it is possible to remove background noise; a (very basic) algorithm could be recreating the foreground by the existing noise removal algorithms and then subtract a certain percentage (your 50% parameter right here) of that from the original audio.

No less impossible then the alteration of individual notes, which, I think, you really oversimplified, TerranRich. Just adjusting the waveforms would eventually result in a chipmunk effect or, of course, considering lower notes, the opposite. Trying to avoid that would be quite easy in the former, but leave you with blanks to fill in the latter case. You'd have to actually recreate the note from the original recording, exactly taking into account where an individual note starts and ends (which stops being simple when the player uses more than just two fingers). Considering the notes were not computer-generated in the first place, they won't be exactly accurate. Plus, you have the player's--let's call it "feeling"--to the song, so you actually have to filter out an individual note from, say, a 5 note chord, alter it, and put it back in again, of course with the original note seperated and removed from the other 4 ones. No more possible than seperating background from foreground.

/tbi