How do I sample from an existing song?

[StefanoEARTH]

Hey all, new to OC ReMix.

I've been making music of various stripes by various electronic means for several years now ( http://www.myspace.com/treesmanufacture for some oldish samples ) - I've used a decent variety of software including ACID Pro 4.0, FruityLoops 3.5, and CoolEdit Pro/Adobe Audition, and I also have a YAMAHA MOTIF7 keyboard synth and some other physical gear.

I've always wanted to do remixes (videogame-related and otherwise) and draw samples from any number of existing songs, but I've never understood how remix artists get clean, distinct samples - do you have to have access to the original recordings so that the instruments/voices are separate, or are there well-established software (or studio) tools and tricks to isolate an instrument (or group of instruments, like drums) in the mix of an existing recording?

Any help will be MUCH appreciated, as I've been trying to research this for quite a while now.

- S.

[Moseph]

Do you have to have access to the original recordings so that the instruments/voices are separate?

This is how official remixes that incorporate elements of the original songs are done.

or are there well-established software (or studio) tools and tricks to isolate an instrument (or group of instruments, like drums) in the mix of an existing recording?

There's a karaoke trick to remove vocals from a song, but the the quality of the results varies from so-so to terrible. Sometimes you can use equalization to cut out a bit of the stuff you don't want, but the only way to really isolate a specific instrument is if you have another recording that is identical except that it LACKS the thing you want to isolate -- in this case, the two recordings can be combined in a way that removes everything but what you want. If you're interested in these techniques, I or another forum member can explain them in more depth, but as far as practical application goes, neither is really very useful.

[Rozovian]

DarkeSword pointed out in some older thread that compression artifacts usually remain after applying an inverted track with just the music on a track with both music and song.

afaik, most professional remix artists work with a studio with the studio recordings of the individual recorded tracks, or at least get the song audio and the full track in high quality so there's far less artifacts when they do it. Note that we're talking big formats here, not mp3s.

[Yoozer]

I've always wanted to do remixes (videogame-related and otherwise) and draw samples from any number of existing songs, but I've never understood how remix artists get clean, distinct samples

The same way you do - use your Motif (or any other synthesizer).

If you hear orchestral drums, there's a good chance they either came from a big sample-pack or via the old-fashioned way - hire session musicians who can play them. Ditto with pretty much everything else. Your Motif is nothing but a big read-only sampler in a box. If you hear a hip-hop beat, take an old drum loop from a 70's funk record and chop it in pieces.

The question you ask is very wide in its scope, but since you use the word samples I can assume you're talking about everything music-related from the Playstation 1 and further (when samples were actually used instead of soundchips).

Any help will be MUCH appreciated, as I've been trying to research this for quite a while now.

Name some examples of which sounds you're looking for

[StefanoEARTH]

Thanks to all three of you guys for coming to my aid - and Moseph, if you or anyone else has time to go into a little more detail on the techniques you described, I'd be very interested, even if it most likely wouldn't turn out to be immediately useful.

And Yoozer, I was talking mainly about the narrower matter of isolating samples from an existing, finished recording, but I am also pretty uninformed, and very curious, regarding a lot of the broader things you mentioned. Where do good sample-packs come from? Most of the voices, drum samples, etc. on the Motif sound pretty good, for example, but most of the samples that I've used in Fruity, Acid, etc. tend to sound too fake or "thin" - lacking something I can't quite put a finger on.

Thanks again guys:D

[Yoozer]

Big sample packs:

Garritan

Kontakt 3

Vienna Symphonic

Spectrasonics

And Yoozer, I was talking mainly about the narrower matter of isolating samples from an existing, finished recording

Yeah, but which ones?

In the hardware world, there's years and years of experience programming samples - often from a limited set of sounds in memory. The Motif's internal memory is much bigger than that of say, an SY35 which was one of the earlier sample-based (well - samples plus FM) synths of Yamaha. A 64 MB piano can sound better than one of 3 gigabytes, simply because of that.

Then there's the matter of effects. Once you start using your Motif in the sequencer mode, with 16 separate tracks, you notice effects dropping away - all 16 tracks have to share the effects. When you play sounds in "Voice" mode, they get 3 effects stacked on them. When you play sounds in "Perform" mode, they have to share effects and suddenly don't sound as good anymore.

A sampler like Kontakt has built-in effects, but generally the idea is that a company that makes a dedicated reverb plugin does the job better than the programmers of the sample library ever can do - so the sample guys leave the work of the effects to someone else. That may be what you describe as lacking something.

Oh yeah, then there's the playing; there are several techniques to let a violin ensemble play (agitato, sforzando, pizzicato, etc.) Not using the same technique for everything helps, too - but this requires quite a bit of study.

[dannthr]

Find a record of this song:

http://www.youtube.com/watch?v=UdaHCLlBkWU

And then add some beatz with another record--then become a famous white rapper!

[Moseph]

Thanks to all three of you guys for coming to my aid - and Moseph, if you or anyone else has time to go into a little more detail on the techniques you described, I'd be very interested, even if it most likely wouldn't turn out to be immediately useful.

A sound wave, in the physical sense, is a sequence of high- and low-pressure that travels through the air from a sound source, much like the ripples that are created by throwing a stone in a pond. If you overlap two sounds which are exactly the same, except that in one the areas of high- and low-pressure are reversed (in technical terms, the phase is reversed), the sounds will cancel each other and you'll have silence.

This canceling phenomenon is used in both of the tricks I mentioned. For the karaoke voice-removal, you need to pan both the left and the right channel of your recoding to the center (in effect creating a mono track out of a stereo track) and reverse the phase on one of the channels. In most cases, this will remove most of the lead vocals. Traditional mixing practice places the lead vocals in the center of the stereo field, so the vocals are equally present in both the left and right channels. When you mash the channels together and reverse the phase on one of them, you are performing the canceling operation on whatever sounds exist in both channels.

The other trick works similarly. If you have two recordings that are mostly identical, playing them together and reversing the phase on one will cancel whatever exists in both recordings but will leave whatever is unique to one of them. So if you have, say, a recording with just drums and a second recording with the same drums and an added bass, doing this will cause the drums to cancel, leaving only the bass.

[Incronaut]

or via the old-fashioned way - hire session musicians who can play them.

Do people on OCR really do that? I never really thought of that, and in some cases i can see it as kinda awkward... but i think thats only cause i imagine a mariachi (or however you spell it) band in my house playing while my mom comes in saying "does your friends want to stay for dinner?!"