MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

‘Extract’ audio from a mix
MUFF WIGGLER Forum Index -> Production Techniques  
Author ‘Extract’ audio from a mix
Northward
Hi
I run Ableton Live and own iZotope Ozone and Neutron.

From a recording session with a band we got a full mix. Not at all happy with the results I’m attempting to make my own mix. So I’ve got all the dry files from Logic. But in spite of asking several times the grumpy studio technician (aren’t they all..) refused to give me the files gone through the hardware/software signal chain. Where the vocals are good sounding.

Q: how do I extract the vocal track transients from the master, by using the dry file? A sort of masking and substracting. I have no idea how I do this with Simpler or izotope plugs.
slumberjack
if this might help?

https://www.soundonsound.com/techniques/extract-vocals-stereo-mix
memes_33
there are a bunch of techniques, but none of them work well. mid-side processing can isolate vocals if they are in the dead center of the l/r (i can't remember if you then need to invert that and mix with un-inverted signal). sometimes i've been able to invert the left signal, mix with right, and then invert the stereo signal and mix them together with either right or left to get something useable. different combinations of inversion and mixing left/right signals can yield different results.

i used to try to extract vocals for remixing purposes, and was sometimes successful with some of the strategies above, but it was never a clean result.

good luck!
Northward
slumberjack wrote:
if this might help?

https://www.soundonsound.com/techniques/extract-vocals-stereo-mix


Great!
Thanks a lot, That was very accurate help
Northward
Double posting.
Northward
slumberjack wrote:
if this might help?

https://www.soundonsound.com/techniques/extract-vocals-stereo-mix


Sorry for these repeated posts, but the forum seems to mess up things today. Not being able to delete is a real helpful feature. The good ol ‘ here’s nothing as a free lunch’ I guess..
Northward
memes_33 wrote:
there are a bunch of techniques, but none of them work well. mid-side processing can isolate vocals if they are in the dead center of the l/r (i can't remember if you then need to invert that and mix with un-inverted signal). sometimes i've been able to invert the left signal, mix with right, and then invert the stereo signal and mix them together with either right or left to get something useable. different combinations of inversion and mixing left/right signals can yield different results.

i used to try to extract vocals for remixing purposes, and was sometimes successful with some of the strategies above, but it was never a clean result.

good luck!


There are these French dudes that does some remixes with some impressive vocal extractions. I’ve been curious about their techniques and gear.

https://soundcloud.com/petit-remix-entre-amis/monkey-maer-the-commodor es-easy-1

Izotope RX is pretty hyped. I’ve worked with an early generation of this program and I’ve was far from impressed. But things gets better..
dequalsrxt
I've messed around a bit with this program a couple months ago - https://audionamix.com/technology/adx-trax/. It needs a high speed internet connection, since it does most of the processing in the cloud.

I was able to get fairly useable results in a short time after first trying it. You basically trace the vocals in a spectrograph, bounce it to the cloud, it spits back rough copies of the instrument and vocal tracks, then you use some tools to clean the tracks up even more, send it back to the cloud to process the final copies. It's 'non-destructive' so if you play the instrument and vocal tracks on top of each other, they fit back together without any seams.

It's definitely not a 'push the button, get good results' sort of thing, you need to get skilled at using the program to the most out of it, which I am not. For example, if there's vibrato in the vocals, you need to trace it otherwise the results will sound flat because you missed information. And it's not always possible to see these things in the spectrograph, so it can be hard to translate from your ears to the screen without practice. One track I tried it on, the vocalist sounded like she had a cold in a couple of parts because I wasn't able to trace her vocals close enough.

You can actually 'rent' the program for a month or however long, which is pretty much why I decided to try it. (Remember to cancel the recurring payment though!) I used the cheapest version, which is $20 a month.

They also have a brand new program that's supposed to automatically spit out stems. It's pretty cheap, $99, but I have no idea how well it actually works and there's no demo videos out yet. I'm skeptical of the 'automatic' part, since it takes some work to get usable results out of Trax.
cretaceousear
I don't understand..
If you have the dry files / tracks / stems just start from scratch. Then do your best to match what the grumpy engineer did on the vocals. Surely that's going to give much better results.
decklyn
Google 'diy acapella'
You can phase invert the instrumental and play it with the full track to isolate the vocals. Many of the remixes on my soundcloud were done via this method. (Link in sig)
There can be some artifacts but amazingly this works.
MUFF WIGGLER Forum Index -> Production Techniques  
Page 1 of 1
Powered by phpBB © phpBB Group