A common question I get asked is, how do I go about prepping stems for a remix? For the uninitiated, stems (or separates) are the recorded component parts of a song, that an engineer will use to create a mix, and that a remixer will chop up and shmangle to create a remix.
I’m writing this article to partly to answer this question, and partly so that I can just link here next time I get asked. Came here via a link? Hello! I’m lazy.

Pre-prep

The first thing you need to consider is who the intended recipient is. Are you preparing a stems pack for general distribution, or for a specific person? Work this out before you start.
If you’re preparing stems for an individual (such as, for example, me), it’s best to start by asking them what parts they want, and in what format. A specific remixer may not want access to every sound in your original track, because they’ll likely be building a new arrangement using their own sound choices.
They’ll want the vocal, pretty definitely, but not necessarily your synth parts, drum programming or guitars. By asking ahead of time, you can save yourself a lot of work.
But whether you’re exporting everything or just some choice slices, there are a few general steps to follow.

1. Resave your project

Start by saving your original project under a new name. We’re going to trash it in the process of exporting these stems, and it would be a shame if you accidentally saved over the original.
Creating a copy with a new name protects you from reflexive saving (I do CTRL+S in my sleep).

2. Make a dump folder

Create a folder on your system to store your stems, and label it not just with the title of the track, but the Beats Per Minute (BPM) – the speed of the track. This is crucial information, and if you fail to mention it, you’re creating a lot of work for your intended recipient. Knowing the original BPM is vital because remixers will use this info when working out how much they need to speed up or slow down recordings by, in cases where they’re changing the tempo.
In fact, the BPM is so important that I’d recommend making it a part of each individual file name.

3. Export the MIDI data

If your original song contains lots of synth parts, do the remixer a favour and export your arrangement as a MIDI file. Most DAWs will allow you to do this easily, it’s usually under the File menu.
Why go to this trouble? Why not just send your original project data? Well, you can, but are you sure that they person you’re sending your stems to uses the same software you do? MIDI is a universal format, and will allow them to import the parts they want, regardless of what VSTs you’ve used.
Even if you’re both using the same platform, having the MIDI file is a boon because most DAWs allow you to import just one part from a MIDI file into a project that you’ve got open (well, FL Studio does, and it’s what I use most of the time). This saves you having to fiddle about with multiple open projects.

4. Remove all FX

Seriously, kill everything. Especially that Autotune plugin. You think you need it more than I do, so in all likelihood you’ve used it with far more severe settings than I would. Moreover, I’ve been using it for a lot longer than you have, and I can get a more natural sound out of it than you can. Please, let me do my job. It’s why you’re paying me.
The intention with a stems pack is to create possibilities, and allow the remixer to do whatever they get inspired to do. If you leave your FX in place, you ensure that your recordings can only be used in the original context you had them in. Why bother with a remix at all, then?
Remove all reverb, all delay, all EQ, even if it makes the recording sound like crap. The more raw the recording, the more the remixer can do with it.

5. Solo and export

For each track the remixer has asked for, do the following. Solo the track, and bring its level up to where it’s about 3db below peaking (ie loud, but leave some headroom). You might have had the track softer in the context of your mix, but we need it clean and loud on export.
Play it back to be sure it’s not running through a pumping sidechain effect or anything that prevents it from being pure raw recording. Then, with the track on solo, use whatever function your DAW has to export it. Make sure you start each export from the same point in the timeline – that way they’ll line up properly on the receiving end.
Save each export in your dump folder, with a name that describes what’s in it. Bonus points for remembering to include the BPM.

6. Protip: Panning for Great Justice

You can save a huge amount of time when exporting mono elements by panning them hard left and right. For example, suppose you want to export two separate vocals. Raw vocals are mono, so there’s really no point in saving them to a stereo file.
Pan one vocal hard left, and the other hard right. Solo them both, and export as a single stereo file, making sure to label it properly. Twice the work in half the time.

7. Zip it up

When you’re done exporting, create an archive of the dump folder. Depending on the file format you’ve used, you might save some space, but most importantly, you’ve collected all of the stems into a single file, making transmission over the Internet a lot easier. In my experience, the RAR archive format does especially well with WAV files.
And you’re done!

A note about file formats

What file format should you be saving in? If you can, ask the remixer. Based on the audio content and what they want to do with it, they’ll probably specify either 24bit 44.1khz WAV, or 320k mp3. Surprised?
While the 24bit WAV is the de facto standard, 320k mp3 is actually completely acceptable for a lot of applications (but not all). Mp3 is undeniably a lossy format, but the artefacts caused by those losses are really only noticeable at lower bitrates, making it perfectly adequate for sending stems over the Internet. Your mileage may, of course, vary.