arottenbit wrote:
magic's everywhere in this, bitch.
http://www.cracked.com/blog/learn-your- … h-the-icp/ XD
Last edited by BitPop (Jul 7, 2010 12:03 pm)
Offline
Using phase cancellation to remove the entire mono center of a stereo mix will usually remove the majority of the lead vocal, but you might also end up losing the bass guitar/snare/kick/whatever else was mono.
using sum and difference techniques, such as the ones being exhibited by the first example in the ableton vid, is a similar process of decoding a stereo signal into it's respective parts (the difference being what is left over after the mono summed information in the center of the stereo mix have been nulled).
so while those processes do kinda work, they are less than ideal because what you are left with sounds very empty (not simply as if the vocal has been sucked out).
trying to isolate or remove an instrument/vocal from a stereo mixdown is a serious challenge, even with all of the processing power we have at our disposal these days.
When the Beatles catalog was remastered last year, they were simultaneously trying to prepare the tracks for 'Beatles Rock Band'. They needed to have each instrument isolated so that when a player messed up on the guitar or drum part, the guitar or drums would drop out of the mix respectively. They were able to completely isolate tracks by using CEDAR restoration tools. CEDAR is usually used in film and archival remastering to remove clicks/pops/hiss/rumble as well as other unwanted background noise. They realized it was powerful enough to remove entire instruments from mixdowns of the Beatles catalog.
check out this link: http://www.cedar-audio.com/news/ibs_lecture2009.html
definitely overkill for doing a lil' remix, but in an ideal world, that's how it would be done.
Offline
this could actually be a good solution: http://www.izotope.com/tech/vocalremover/
i haven't tried this specific piece of software, but i have used izotope's RX noise reduction software, and that has worked wonders for me in the past with restoring poorly recorded audio.
if they have figured out how to optimize their RX software for specifically removing lead vox, this may indeed work pretty well. the examples in the sample playlist are pretty amazing (granted, they are the ones who chose those specific clips).
this is why i think this may be a better solution:
"The iZotope VocalRemover uses an advanced set of filterbank signal processing algorithms to remove the vocals while still preserving the stereo image of the music. Many simple products use "center channel removal" which ends up removing a lot more than just the lead vocal."
much better approach.