MattO
Newbie
Joined: 31 May 2005
Posts: 9
|
| Re: CSP vocal removing |
 |
|
quote:
Originally posted by MattO
quote:
Originally posted by wailingalleycat
The problem with that site, it doesnt say how they do it.
One method which works on some stereo material is based on FFT like a lot of recent DSP audio stuff.
The audio is split into, say, 8192 frequencies, about 50 times per sec. Then individual frequencies are suppressed on the bases of left-right differences, phase and/or amplitude. So the vocal is dynamically notched out on a very fine grid. It's not perfect but works fairly well in a number of cases.
There is now a full article, find out how Vocal Removal can be done
http://www.csp-audio.com/vocalremoval_article.htm
|
Thu Jun 16, 2005 2:36 pm |
|
|
conquistadore
Forum Moderator
Joined: 03 Aug 2004
Posts: 541
|
| FFT and DFT |
|
|
Thanks MattO, that did quite explain it..
Could you expand a bit on DFT and FFT? |
|
Fri Jun 17, 2005 12:27 am |
|
|
|
|
MattO
Newbie
Joined: 31 May 2005
Posts: 9
|
| Re: FFT and DFT |
|
|
quote:
Originally posted by conquistadore
Thanks MattO, that did quite explain it..
Could you expand a bit on DFT and FFT?
You can view the DFT as a huge filter bank of FIR filters, splitting the sampled signal into all its frequencies. And because the signal is finite in time, and sampled, it can be represented by a finite amount of frequencies. Therefore you can also transform all the frequencies back to the original signal in time.
The FFT is a fast (but exact) version of the DFT, that is useful for many things. It is also applied extensively in convolution reverbs.
If you really want to know more, here is a nice link.
http://ccrma-www.stanford.edu/~jos/r320/
MATHEMATICS OF THE
DISCRETE FOURIER TRANSFORM (DFT)
WITH MUSIC AND AUDIO APPLICATIONS |
|
Sat Jun 18, 2005 12:04 pm |
|
|
raykeefe
Newbie
Joined: 16 Sep 2002
Posts: 8
Location: Melbourne, Australia |
| vocal removal technique |
|
|
The phase inversion technique is the main one I use but it only works on dead centre vocals. If the vocal has a delay to one channel then that creates problems.
I use Logic Audio on PC still and have set it up with one send bus having EQ either side of an Analog X vocal remover plugin. A second send buss allows me to add back the bass and drums that were removed. Usually the bottom end is also in phase and gets removed.
This works fairly well on some tracks but the use of stereo reverbs means you can get some of that left over. The ghost vocal effect.
There are more sophisticated techniques available including FFT (already explained) and I have heard of a unit that handles vocals with arbitrary delay and phase effects channel to channel.
For some musicals where there are a lot of singers it gets pretty hard. EQ can be used but you also lose a lot of the backing track. For these I sometimes add back a few parts using midi instruments to fill out the backing track again.
_________________
Ray |
|
Mon Jun 20, 2005 6:24 pm |
|
|
|
|
 
|
|
Goto page Previous 1, 2
All times are GMT. The time now is Fri Jul 30, 2010 12:58 am
|
|
|
|
| |
