What I want to do is have a different person replace the vocals of another song/recording. Basically, I have this Weird Al song, but I was wondering if I could change his voice pattern to make to sound like someone else (like say, someone from the Beatles or someone famous) singing it. Is that possible? Please help and tell me if there’s a way.
You could try the “Center pan Remover” (often called Vocal Remover) effect to try and remove the original vocal (may or may not work - it depends on the original recording), http://audacityteam.org/download/nyquistplugins and then adding a replacement vocal
supermariostar,
Almost certainly not. At least, not in the way that you’re likely thinking.
If you have access to the original multi-track recording from Weird Al and access to one of the beatles then it would be easy. But if you don’t then it’s not going to happen.
No hope.
The Center Pan Remover takes the original voice out according to some very strict rules, and then destroys it. If everything goes well, it also leaves you with a mono orchestration. If things don’t go well, you get a mono orchestra track and a bubbly, distorted voice.
You still need the original mix tapes to do what you want.
Koz
Alright, I’ll try with the center pan remover and see what happens. Although, how do I put the center pan remover in my effects section? (it has .ny at the end when I downloaded the zip, it is different then the others, how do i put it in there?)
Didn’t it come with a ReadMe? I think mine did.
In Windows, it’s Program Files > Audacity > Nyquist or Plugins.
In Mac, it’s the same two folders wherever you put them.
http://audacityteam.org/wiki/index.php?title=Vocal_Removal
Here’s way to do it without the tool.
http://audacityteam.org/help/faq?s=editing&i=remove-vocals
Koz
Oh, one more. Sometimes you have to restart Audacity for it to catch on to what you did.
Koz
Do you more or less understand what all these tools are doing? Any singer in the exact middle of a stereo show has electrical waveforms that are identical on both left and right. If you cause one side, say Right to be inverted and then smash Right and Left together, the singer will vanish.
I made a sound clip that illustrates this. Download “Left-Right” from here. It’s 38 seconds.
http://www.kozco.com/tech/soundtests.html
If you magnify the “center performance” waveforms a lot (segment three), you will see that the wiggles are exactly the same on left and right. The last segment I engineered my voice to be “flipped” on the Right channel. This is an engineering test, so there’s no guitar or anything back there. Blow that up and look.
If you make a mono track out of this clip or listen on a mono mix speaker, my voice will vanish in the last segment.
The important point is to notice that any instrument not in the middle will survive. My voice on segments one and two goes right through.
The problem is what happens when my voice is not in the middle. I won’t go away with this process. Unless the singers are in the exact middle, you lose. If the singer’s voice is in the middle, but their echo or special effects isn’t, only the special effects will remain. This gives you the singer performing out in the parking lot or down a long hallway.
If the drums are in the middle (usually), they go, too. This isn’t as easy as you would think.
Koz
Thanks, Well, I’m able to get rid of the vocals to make it sound sorta like an instrumental, but now is there any way to change the voice pattern to make it sound like someone else is singing the same song? if possible, how would i do that?
<<>>
The down side of all these tools is that they don’t take the voice out and put it somewhere. They take it out by destroying it. They work by vocal cancellation.
You can play the voiceless orchestration back into headphones and then sing or ask somebody else to sing along. Record that using your local microphone and Hardware or Software Playthrough for cuing. Once you have the new voice on its own track, then you can throw all the effects at it and change it around to do what you want. Do a mixdown in Audacity to integrate it with the original show and export as a new song.
Because of latency problems, you may also have to slide the orchestra or voice sooner or later a little to match it up.
Actually, what I would be doing is capture the new live performance before I took the original vocal out. That’s much better for cuing.
Koz
is there any way to change the voice pattern to make it sound like someone else is singing the same song?
Aside from re-performing it as Koz mentioned, no.
In order to do that you’d need two things:
- a solo voice track.
- a plugin to change one voice from another.
#1 you can get using the Center Pan Remover and just subtracting the original signal from the signal put out by that plugin. But #2 doesn’t exist, the human voice is too heavily filtered by your lungs + throat + mouth to re-filter it and make it sound like someone else. It’s even more impossible to make it sound like someone specific.
Only slightly off topic:
There are now “virtual singers” 
http://www.zero-g.co.uk/index.cfm?articleid=805
<<<#2 doesn’t exist>>>
No, but I can think of ways of doing it. The tools are certainly there. Given that you have the voice track by itself, apply a nice FFT with the ability to manipulate the Fourier elements independently from each other. They have equipment that can take a bad answering machine and tell you with absolute certainty who it is with similar tools.
I imagine a nice engineering student should be able to handle that before lunch.
Koz
Koz,
You’re talking about shifting the formats in someone’s “voice pattern” to a different set of formats (you’d also likely need to change the fundamental frequency, but that’s trivial). You can do things like that, it isn’t very hard (it’s basically just a very radical EQ). But there’s more to a “voice pattern” than just frequency response. In order to be convincing, you’d also have to change the timing, pronunciation, accent, and breath noises. I don’t doubt that it’s technically possible, but it’s still a ways off.
<<<I don’t doubt that it’s technically possible, but it’s still a ways off.>>>
So you’re not going to whip this out before lunch?
I’m crushed.
Koz
Heh, well I’ve been fooling around with Audacity, but can’t seem to do it yet. Koz, when you were talking about the FFT, did u mean that it is possible to mess around with the voice pattern or something using that?
Can you actually change someone’s voice with this?
No, but you can synthesize a “replacement” vocal.
It’s actually a hybrid sampler/synthesizer - It sounds a bit strange (synthetic) as a lead vocal, but it’s pretty good for backing vocals. Also, it has the advantage that it will “sing” for you any time of the day or night, sings in tune, does not have tantrums, and won’t run off with the drummer 
<<<Can you actually change someone’s voice with this?>>>
No. That was a not so subtle joke to get Latham to write a tool that would do that.
He didn’t take the bait…yet.
Koz
Ok well, I’m just still messing with it, but nothing. I’m trying to somehow access the voice pattern, but no luck. i.e., I’m trying to make Weird Al sing “Mind Games” which is by John Lennon, and basically, I want Mind Games to sound the exact same way , just with a different voice singing it. Is there any effect or something to do that? n I’m trying to make this go possible for any song, (if it is possible now).
An effect to change John Lennon into Weird Al ?
No there isn’t (thankfully) 