Mods, move this to Café Society if I’ve misplaced it.
Alright, so I love mash-ups. What’s that? Never heard of them? Smells Like Bootylicious is a great example: the guitar riffs from Smells Like Teen Spirit with the vocals from Bootylicious, remixed together to make a song that’s funkier and cooler than either of the originals. Another good one is Oops! The Real Slim Shady Did It, where Eminem bashes Britney, accompanied by the groove to her single Oops!.. I Did It Again instead of his original (kind of boring) hook.
So, I wanted to do a few of my own. My first candidate is to take samples of The Funniest Answering Machine Message in the World and remix it with Kiss the Girl or Part of Your World from Disney’s “The Little Mermaid,” and then add in a dance beat. I’ve got a bunch more that I think would be great: Garbage’s Cherry Lips and Nikki Costa’s Everybody’s Got Their Something, for example.
Anyway, I wanted to do these on my PC, at low cost. I tried Googling, but I honestly don’t know what terms to use. I want to separate MP3 or WAV files into vocal and audio tracks, but I can’t figure out how a computer would even figure that out, let alone what software would do it. So, here are my general questions:
Is it possible for a computer to take a final, one-track mix (like an MP3) and break it back into its component tracks (bass line, melody, percussion, vocals), or is it like Photoshop, where once it becomes a JPG, George Bush’s head is always going to be on Pamela Anderson’s body?
Is there inexpensive software that will let me do this to a pair of WAVs or MP3s and then down-mix it into one track, so I can put it on a mix CD?
Is it possible to “un-mix” a track?
In a word, no. But you can use an equalizer to filter isolate a frequency range. You could, for instance, pull out a bassline and by filtering evrything above 200 Hz or thereabouts. But if there’s a kick drum on the same track you’ll never be able to separate the kick drum from the bass guitar.
The problem is that most instruments have a fairly broad frequency range that overlaps with other instruments.
There’s another trick for isolating vocals. Usually vocals are centered between the left and the right channel. If you separate the channels, invert one, and add them back together, the vocals will be significantly reduced. Along with any other instrument that is centered in the mix. There’s a utility that does this automatically. Vocal remover by Analog-X. It’s a plug-in. you need a host program like Cakewalk, Cubase, or Adobe Audition to use it.
There’s a very capable audio editor called Audacity. I don’t think it hosts direct-x plug-ins, so you couldn’t use the vocal remover with it. You could, however, do it the manual way: separate the left and right channels into two wave files, invert one, and mix them together as a single mono file.
Rather than trying to separate the instruments, you might have better luck just finding a section of the song that has the parts you want, and looping it. Sections like an intro, chorus, or break.
If you’re interesting in working with loops a lot, try Sonic Foundry’s Acid. I think there’s a version for $70 that will let you work with up to eight tracks at once. You can work with loops in Audacity, but Acid makes it a lot easier.
A good source for freeware and shareware music is the Shareware Music Machine. (They do have some popup ads, though. You might find a free editor that hosts direct-x plugins, or a stand alone vocal remover utility.
Oh and finally, if you’re going to use your mixes and mash-ups for any commercial purposes, or play them live, you need to clear the copyrights. Kind of goes without saying, but I want to be clear that I’m not encouraging any illegal activity. Just covering my butt and keeping this message board out of trouble.
Wow, thanks! Amazing what you can learn from a simple question. One more follow-up for you, heresiarch (or anyone else) - by inverting one track and adding them (sounds like an AND function), I can remove the vocals; is there a similar “OR” function that will let me keep only the vocals? This would be really helpful.
I couldn’t think of a way to do an XOR operation directly, but I did find that some audio editors can convert left/right stereo tracks into a middle/side split. Maybe they use an XOR operation.
Most of the vocals would be in the middle (along with bass, but you could filter most of the bass out with a high pass filter).
I also found that the new version of Adobe Audition (which was Cool Edit Pro before Adobe bought it) has a feature specifically for either removing vocals or for making a capella versions. The program costs about $300, but there’s a demo version available.
Longtime user of both Sonic Foundry’s Acid and CoolEdit Pro/Adobe Audition here.
The vocal cut and a capella tools heresiarch mentions in CoolEdit/Audition just use the same phase cancellation algorithm he already described in his first post. (The a capella would invert the vocal cut, & sum it back to the original track). You’ll rarely if ever get the results you’re looking for when using it, even with very careful tweaking and frequency filtering. Only in the highly unrealistic circumstance where what you want to cut is centered, and what you want to keep is panned out wide, and the parts you want to keep stay out of the frequency range of what you want to cut will you get results like what you are looking for. CoolEdit/Audition is a great program, but don’t buy it expecting magic from this specific feature.
Sonic Foundry’s Acid is a very cool and very fun program, designed for creating loop-based music. If you have interest in creating dance/electronic/pop loop-based music, you’ll have lots of fun with it, and there’s tons of royalty-free loops out there to play with. I’ve gotten way more hours of entertainment out of Acid than I have from any computer game I’ve ever bought. Also, for quite awhile, pretty much anything you boughy from Sonic Foundry came with an offer for a free copy of Sound Forge, which is their basic wave editor (a pretty healthy subset of the functionality offered by CoolEdit/Audition).
