making my own instumental
- Justin_topia
- Joined: Mon Jul 28, 2003 5:47 am
- Location: Northern California
making my own instumental
I have a song that i want to extract the vocals from. I have already searched for an instrumental version, but it does not exsist. I was hoping that somebody could please give me some help in finding a software that will exstract them for me.
- madmag9999
- Joined: Sun Aug 10, 2003 11:50 pm
- Status: Engaged
- Location: Pennsylvania
u might have truble doing this becouse u need a kareoke program and iv tried to look for one a while ago but it seems that its illeagel to do it w/o a kareoke machin or something like that. i suggest u search download.com for a kareoke program. u could also look on kazaa or something but like i said i could find anything that can do this.
Moonslayer's Guide to a-m-v.org | AD & ErMaC's Guides to Audio & Video
"I'm sorry but i don't trust anything that bleeds for 5 days and doesn't die."
"I'm sorry but i don't trust anything that bleeds for 5 days and doesn't die."
- Farlo
- expectations of deliberate annihilation
- Joined: Tue Apr 02, 2002 8:04 am
- Status: The Dark Host
- Location: Fort Smith, Arkansas
- Contact:
- Lyrs
- Joined: Thu Aug 29, 2002 2:41 pm
- Location: Internet Donation: 5814 Posts
- koronoru
- Joined: Mon Oct 21, 2002 10:03 am
- Location: Waterloo, Ontario
On some songs, depending on how the music was mixed, you can get pretty good results by computing the numerical difference between the left and right channels. In Cinelerra I'd do that by panning both channels to the centre and applying the "invert audio" plugin to one of them. On other songs, the result sounds like crap; and even if this technique works, it causes you lose to all the stereo information. Try it with your particular song and see if you can tolerate the results.
You can get commercial hardware or software to remove vocals. It is expensive. Some of it works by implementing the "subtract" technique mentioned above, and it works exactly as well as that technique works in any other implementation. Karaoke machines generally use the subtraction technique, and/or special recordings which (whoa, innovative idea here!) actually don't contain vocals in the first place. The fancier vocal removal boxes that the pros use (when absolutely forced to do so - no pro will be happy about using such a box) use special DSP algorithms that attempt to recognize the spectral profile of vocals. Some of those work pretty well. None of them are as good as you'd really like them to be.
There is no ultimately good way to do it because it is, fundamentally, a difficult problem. A vocal remover would have to analyse the sound, figure out which things in the sound "sound like" voice and which "sound like" instruments, and then separate them. Only a human brain can really do that, and even a human brain can't really get it right 100% of the time. Just try listening to one of those songs from when vocoders were cool - my favourite example being "Crimson and Clover" by Tommy James and the Shondels. There are points in that song where the electric guitar sound is used to pronounce understandable words (they ran the guitar and a mic into a vocoder). What should a vocal removal filter do with that? Even assuming it was able to separate out the vocoded-guitar sound in the first place (which would be a major feat) it's not at all clear whether it then ought to consider it a "vocal" or an "instrument".
You can get commercial hardware or software to remove vocals. It is expensive. Some of it works by implementing the "subtract" technique mentioned above, and it works exactly as well as that technique works in any other implementation. Karaoke machines generally use the subtraction technique, and/or special recordings which (whoa, innovative idea here!) actually don't contain vocals in the first place. The fancier vocal removal boxes that the pros use (when absolutely forced to do so - no pro will be happy about using such a box) use special DSP algorithms that attempt to recognize the spectral profile of vocals. Some of those work pretty well. None of them are as good as you'd really like them to be.
There is no ultimately good way to do it because it is, fundamentally, a difficult problem. A vocal remover would have to analyse the sound, figure out which things in the sound "sound like" voice and which "sound like" instruments, and then separate them. Only a human brain can really do that, and even a human brain can't really get it right 100% of the time. Just try listening to one of those songs from when vocoders were cool - my favourite example being "Crimson and Clover" by Tommy James and the Shondels. There are points in that song where the electric guitar sound is used to pronounce understandable words (they ran the guitar and a mic into a vocoder). What should a vocal removal filter do with that? Even assuming it was able to separate out the vocoded-guitar sound in the first place (which would be a major feat) it's not at all clear whether it then ought to consider it a "vocal" or an "instrument".
