How to Prepare Stems for Mixing: An Engineer's Checklist
I've been mixing records for a long time, and here's something that surprises a lot of artists: the mix starts before I ever touch a fader. It starts the moment you hit "export" in your session. Clean, well-organized files mean I spend my time on the song: the vocal, the groove, the emotion. Messy files mean the first hour goes to detective work.
The good news is that preparing stems isn't hard. It's a handful of habits, and once you've done it right one time, it becomes routine. Here's the checklist I'd give any artist, band, or producer before sending tracks out to be mixed, whether that's to me or anyone else.
First, what do we mean by "stems"?
People use the word loosely. Strictly speaking, stems are grouped submixes (all the drums on one stereo file, all the backing vocals on another), while multitracks are every individual track on its own. For mixing, I almost always want the individual tracks: kick, snare, each guitar, lead vocal, each harmony. The more separated things are, the more I can do.
If you've already committed to grouped stems for a reason, that's fine. Just tell your engineer what you're sending.
Start every file at the same spot
This is the big one. Every file should start at the exact same point, usually bar 1, beat 1, or the very top of the session, and run to the same end point. Even if the tambourine doesn't come in until the second chorus, export it from the top so it has silence at the start.
When every file starts at the same point, I drop them all into my session and they line up perfectly. When they don't, I'm sliding parts around trying to guess where the bridge vocal was supposed to land. Nobody wants that, least of all you.
Also let reverb and delay tails ring out at the end. Don't chop the last note off.
Match your session's format
File type: WAV (or AIFF). Never MP3 or other compressed formats for mix files. Those throw away information you can't get back.
Sample rate: Whatever you recorded at. If your session is 48 kHz, export at 48 kHz. Don't convert on the way out.
Bit depth: 24-bit is the safe standard. Some engineers are happy with 32-bit float. If you're not sure, ask.
Mono vs stereo: A mono source (a vocal, a bass DI, a single mic on a guitar amp) should be a mono file. True stereo sources (stereo keys, a stereo synth, drum overheads) stay stereo.
Normalization: Off. Leave the levels as they are in your session.
Pull the processing off your master bus
If you've been working with a limiter or "loudness" chain on your master bus, that's fine for your rough mix. But bypass it before you export the mix files. Mastering is where final loudness gets handled. Printing it into the stems just ties the mixer's hands.
The same goes for clipping. Check that no file is hitting 0 dBFS or going into the red. If a track is clipping, pull its fader down and re-export. You don't need to hit an exact number; just leave comfortable headroom.
(If you're wondering how loud the finished master should be for streaming, that's a whole post of its own, and it's coming. Short version: Spotify turns tracks up or down during playback to a target of about -14 LUFS, so crushing your files for loudness doesn't buy you what it used to.)
Effects: print it, or leave it dry?
My rule of thumb:
If an effect is part of the sound, like that amp sim you built the guitar part around, a filtered vocal in the intro, or a synth patch with its own delay, print it. It's a creative decision, and I want to hear what you heard.
Time-based effects on the main elements (reverb and delay on the lead vocal, for instance) are usually best sent dry, so I can build the space to fit the whole mix. If you love your reverb, send both a dry version and a wet version, clearly labeled.
Vocal tuning: if tuning is already done and part of the performance you signed off on, print it and say so. If you want tuning handled as part of the mix, send the untuned vocal.
Name files so a stranger could follow them
"Audio 1_03.wav" tells me nothing. "14_LeadVox_Chorus.wav" tells me everything. A few tips:
Number the files in a logical order: drums, bass, guitars, keys, vocals.
Use plain names: Kick_In, Kick_Out, Snare_Top, Gtr_L_Rhythm, BGV_High, LeadVox.
Put everything in one folder, zip it, and upload it as one file.
Send the "why," not just the "what"
The files are half the package. The other half is context. Include:
A rough mix (your current bounce, limiter and all). It shows me the vibe and balance you've been living with.
Tempo, key, and sample rate.
One to three reference tracks you'd like the mix to live near, and what you like about each (the vocal sound, the low end, the overall energy).
Short notes: "The chorus needs to lift," "Keep the drums roomy," "The bridge vocal is the emotional peak." A few honest sentences go a long way.
Before you hit send: a quick checklist
All files start at the same point and run full length
WAV/AIFF, session sample rate, 24-bit (or what your engineer asked for)
Master bus limiter and normalization off; nothing clipping
Effects decisions made (printed vs dry, labeled)
Files clearly named and numbered, in one zipped folder
Rough mix, tempo, key, references, and notes included
Do those six things and you've done your part to get a mix that sounds professional. The engineer can focus on making the song feel the way it should.
The same "clean files, clear notes" idea applies to spoken word, by the way. If you're a podcaster or an author working on an audiobook, the prep looks different, and I cover that side of things on the audiobooks and voice-over page.
Ready when you are
If you've got a song that's ready for a proper mix and master, I'd be glad to hear it. You can see how remote mixing works with me on the Online Mixing page. Send your stems, tell me what you're going for, and we'll take it from there. And if you're not sure your files are ready, ask. I'd rather answer a question up front than have you re-export later.
— Jim
Comments