Bleed all over!

AlanRatcliffe

...the backing vocal tracks that is.

I've been helping to mix a CD for an artist. Some great stuff, but the tracking (not my work) suffered from an overly present mic, guitar bleed into the vocal mic and, most importantly, a huge amount of headphone bleed into the backing vocal tracks (up to 8-part vocal harmonies!).

With the backing vox I suspect that the two singers who did bulk of the harmonies made the classic error of leaving one can off the ear and the engineer did not send them a mono mix into the can they were using and turn off the exposed can (or simply smack them upside the exposed ear ?). Most of their parts are fairly quiet to begin with, so the bleed comes through loud and proud in each track and the stacked effect in the mix is enough to throw off the phase, panning and EQ of the main vox and instruments.

I was using Cubase SX - not a program I'm very familiar with and this was the first version with nothing but the bundled plugs, so far from ideal. I basically tried to EQ out any unneeded frequencies on each voice and gated as best as I could with the bundled comp/gate. Still, the results were far from ideal.

I've persuaded the artist to send me the source tracks to redo the mix in my studio, where I have software I know backwards and a large selection of quality tools. It also strikes me that I can try reversing the phase on half the tracks - I might have to pan parts in pairs to be most effective, but I figure it's worth a try.

Anyone else got any tricks?

Mo-Facta

Try and use a gate with a range (floor) parameter. It can helps to control the bleed without killing the entire signal when the gate closes. You'll still probably get some top end poking through when the gate opens though.

You could also try expanding or anti-compressing (using a ratio less than one, i.e. .05) and then compressing the peaks back down, sort of how Dolby noise reduction works. Two Waves RComps in series can accomplish this.

Or you could try a dual-band gate, like the MCR Dual Gate VST. It has two bands with independent attack and release controls along with a selectable crossover frequency. Incidentally, it works great on toms. Here it is:

http://thewombforums.com/attachment.php?attachmentid=9665&d=1289245582

Other than all that, I don't have any more suggestions besides a spectral fft editor or manual editing out of the extraneous noise.

Cheers ?

AlanRatcliffe

Yeah, that's what I figured, but was hoping someone had a magic spell. You never know... ?

The problem with any dynamics processor is that the vox and the bleed are often at the same levels. The singer's head movement also causes phasing, so any kind of spectral reduction system is going to be of limited use.

So it's probably a case of zooming in close and manually editing out everything I can on each track, doing what little I can with EQ and then mixing them a bit softer than I would usually. I'll fiddle with time alignment too, see if I can minimise the problem a little more. I don't mind a little bleed - IMO it actually helps glue a track together, so I don't need to remove it entirely, just get it under control.

It's just frustrating, as if the tracking engineer had taken an extra few minutes to check the takes, he would have realised there was a problem and redone the takes and saved me countless hours of editing.

Mo-Facta

Totally agree. My suggestions were a shot in the dark considering I haven't heard the material.

Wish I could have been more help! Good luck!

Cheers ?

AlanRatcliffe

Thanks! ?

singemonkey

Good cautionary tale this.

Arno-West

Alan, I don't know how much time you've spent so far on trying to fix the problem, but IMO it is sometimes less time-consuming to get the singers to come and redo their parts. That shouldn't take too long since you now know exactly which parts you want and where they are in the song. I personally would rather spend an extra 2 hours re-tracking than waste 8 hours editing parts and end up still not happy with the result.

AlanRatcliffe

Arno West wrote: Alan, I don't know how much time you've spent so far on trying to fix the problem, but IMO it is sometimes less time-consuming to get the singers to come and redo their parts. That shouldn't take too long since you now know exactly which parts you want and where they are in the song. I personally would rather spend an extra 2 hours re-tracking than waste 8 hours editing parts and end up still not happy with the result.

Unfortunately not an option for a number of reasons. Artist is happy with what I've done already, so is prepared to live with it, but I'm not. ?

AlanRatcliffe

OK. Here is the remix. Comments welcome:
https://www.box.com/s/mcdred2js2m4zbiiywdl

Still a bit of a problem with sibilence on the main vox, but I can't fix it better without really making the vocal tonality odd.

V8

Had a quick listen - the overall mix sounds pretty damn fine imho.

If I wasn't listening for it, I wonder if the ffff's and esss's (sibilance) would have been as noticable to my 'ol ears. That said, 1:50-2:20 the backing vocals and main vocal have a fair amount that didn't feel right, although towards the end (last 30s) it seems like a nice effect that I'd be tempted to try with some stereo widening on the backing vocals.

Would be nice to hear the original in comparison?

AlanRatcliffe

Due to be replaced, so you'll need to be quick:
http://www.reverbnation.com/play_now/song_14624347?utm_campaign=opengraph&utm_content=song&utm_medium=link

You'll hear quite a few mistakes in the contrabass part which I've since fixed and some ungodly phasing artifacts. Couldn't fix timing or pitch errors on the voices because of the bleed, so I left the original loose feel fairly intact. I also reamped the DI'd acoustic through my Aura pedal to get rid of most of the zing.

Next two tracks are the same project and I played most of the instruments on those. Trying to step Bill up to a more radio-friendly product on those tracks. They are still being remixed, but weren't as bad as Song for Syd as the instruments were done in my studio and only vocals and one or two guitars were tracked elsewhere.