Audio Question from my Uncle Randy
Author: Gtown Monkey
My Uncle is sort of a tech head,
musician and more recently Halloween Superfreak.
He recently posed this audio question...
"I'm trying to find a way to remove or at least suppress
a voice on a musical track. I had some success by inverting
one of the stereo tracks but not enough.
Would you know of any way to do that?"
any thoughts from the pros here?
(btw, he's PC based)
i think brows is better suited for this ques than me, but your unc is on the right path imo. if he had an isolated version of the vocals (an acapella, for example) he should theoretically be able to cancel it out of the live track.
this ball's in brows (or dr browm's?) court, i believe
Unless he has access to the original multitrack, and "physical" editing won't work to restructure the song, then he's left with phase inversion and EQ..and no guarantees:
The Truth About Vocal Eliminators
"Removing Vocals"
Vocal Removal
Um, not to ask the obvious... but is there any chance that a "karaoke version" of the track in question might already exist? (Or is he trying to create a karaoke version of a fairly obscure song?)
thanks dudes...
Hey anybody heard from Stitch lately?
try myspace. gmail? not so much
here is how i removed the vocals from immigrant song...
first i got a hold of the program audacity... i think they make it for windoze too... if not, try and find some sort of wave editor with advanced noise removal.
so i opened the song in audacity and went about using the noise removal effect to get rid as much non-vocal sound as i could.
noise removal lets you select a range of the waveform and have the effect "learn" those frequencies. you then apply what it has learned over the whole song, and those frequencies are removed. so in this case i was basically selecting from any area with no vocals. after sort of a general removal, i did additional passes of the effect on the non vocal sounds that were still left (sometimes just having it learn a short sound such as a snare or cymbal).
eventually i was left with sort of a crummy sounding acapella. i took this and inverted the waveforms and placed it over top of the original in cubase. i may have had to fuck with eq and stuff a little bit at this point, but for the most part the vocals were gone. i am sure gates and noise suppressors could help at this point as well.
there are other ways like what dr. brown said... for example if a song is made with a sampler and you can find a loop with no vocal, you can invert it and put it under a section that has the same loop plus a vocal. i got an acapella of it could be sweet by portishead in this manner.
be forwarned that these processes are fairly painstaking... i spent hours doing the portishead and the zeppelin.
The one-click solution is to invert one channel and mash em. It will work more than you would think. Some surround sound processors would work that way to fake surround sound on something that was just a standard stereo recording. I think it was intended for movies/TV to add the effect to the non-dialog audio.
That will only work if the vocals/dialog are in mono. often there are many vocal tracks recorded on top of each other, and to give it extra depth they may pan some of them a little more to the left or right. That would explain why he may not be getting the best cancelation.
That cancellation technique can also knock out some mono instrument tracks, but it's surprising how often it just cuts vocals. Back in high school i had a standalone surround sound processor that did this to the satellite speakers. Actually... that thing may still be in my closet somewhere? yikes.