Absolutely! One would think, "easy" until you look at what's involved. If you're lucky enough to find a raw, multi-channel and 'unmixed' version of a song with vocal and musical instruments, you will definitely be able to achieve some wonderful results. Although trying to get your hands on a commercially popular song (as such) through typical Internet methods is damn-near impossible. Maybe we'll all live long enough...

 

​

 

In all seriousness, Sound On Sound magazine this month deals with this question in great detail. Check it out at the sister site.

 

Sony Spectralayers Pro 3 has the tools to isolate vocals and their harmonics. Izotope RX is an alternative, but I'm more familiar with Spectralayers, which is truly a "photoshop for audio". Use in conjunction with Melodyne's polyphonic "DNA" mode for quicker (& possibly better) results. Trial and error.

Before you start, some preparation can get you there faster. Remove the Side content, keeping only the Mid - where vocals are located 99% of the time (tip - don't just collapse to mono; actually remove the side channel). Use a filter to slice out dispensable lower and higher frequencies as well as problem areas. You can also try Zynaptiq's Unmix Drums to get rid of transient/rhythmic content - tread carefully though cause you can easily damage the vocals with excessive settings. These three things should make the job a bit easier.

Honestly, the extraction with Spectralayers is going to be a very cumbersome job... you have to manually select the parts of the spectrum you want to keep, formants & harmonics, and make sure you remove all extraneous junk for a clean result. Don't expect pristine results without a lot of elbow grease, but remember that if you are going to be packing them into a thick mix, you don't have to isolate them with excessively painstaking precision.

Use a gate to cut out quiet bursts of remaining junk. A touch of excitation using things like Sonnox Inflator, Waves Saphira, Aphex Aural Exciter, or a pinch of your favourite saturation can go a long way to restoring some of the lost vocal content esp. in the higher frequencies. Then compress it smoothly using an opto style compressor to even it all out. A recent discovery of mine is Flux IRCAM Trax V3... it can probably help restore some of the natural character to the vocals and/or shape them to your liking. I'm still exploring this, so I can't give you any specific tips, but it's a really crazy tool for vocals.

Wish there was a simple solution, but aside from acquiring the vocal stems from the artist, this is pretty much your best bet. Good luck

 

Last edited: Jun 16, 2016

Call a good musician like there are here and ask him to make the music!!!!!

 

Diamond Cut Forensics10 Audio Laboratory as well. I'm gonna check out the Sony Layers thingie...sounds complicated enough to be intriguing, and given the manual nature of it, the effects can be unique...when applied for manipulating sound by surgery for creative effects, and reading the Sound on Sound magazine article...

 

Last edited: Jun 17, 2016

hmmmm thanks for all your responses. I think the best bet will be to find someone to cover it. it will be very difficult otherwise and I can see my self getting frustrated with it.

Thanks again you all have been great

 

Happy Father's Day

"Frustration" is just a sign of learning with a person that's spoiled...everyone goes thru a time of wtf, I can't do shite period. Ppl that can persist get better, ppl that have a remote control and just change the channel, just watch TV and do nothing...imo.

Dare to be bad!℠


--------------------------------------
*Dare to suck℠ sounded to me, too much like a date of mine...

 

Last edited: Jun 19, 2016

What part do you want to extract? It's possible they've sampled the speech from another documentary. I know I've seen the Michio Kaku speech at the start before. A bit of youtube searching may yield results. Otherwise, I can have a crack at it for ya.

EDIT: Her voice sounds really familiar too. Kinda like Julianne Moore

 

Last edited: Jun 19, 2016

Just tried Forensics10 based on this and it's really impressive, wow

 

1: Take the acapella and use phase cancellation to remove the music.
2: Use Sony Spectralayers Pro.

 

Cool, glad to help.

Happy Father's Day, about to go to dinner, and not be the guy cooking on the bbq on my own freakin day. Misandrist West spank you, spank you alot.

 

Only thing which i've tried which can do this sorta ok is AV MUsic Morpher Gold.

It combines Imaging + formant tracking (I think() to perceptually remove/isolate center panned vocals,.

I wish I could reccomend something less gimmicky with a better UI, but that's the best I know..

 

All good if you want to have a go lol. The part that makes it iffy is the actually quality. Hard to find a nice clean one. Also her speech is pretty one off. No idea where its from. I've tried a few things. I can get it close to where i want. On the other hand im thinking about a vocalist because i wouldn't mind changing a few things to better suit.

 

RipX FTW. Merry Christmas.