Vocal Stem Prep Checklist: Send Vocals to a Mixer the Right Way
Vocal Stem Prep Checklist: Send Vocals to a Mixer the Right Way
Pick your DAW and who the files are going to. Tick items as you go, then copy the finished list into the email with your stems. Every item exists because an engineer has had to send that exact question back to an artist.
Common stem mistakes → what the engineer sees → fix
| Mistake | What the engineer sees | Fix |
|---|---|---|
| Files start where the first region starts | Lead lands bars early; every file needs nudging by ear, and the adlibs never quite sit | Export every file from bar 1 (or one shared start) so they're all the same length |
| Reverb or delay printed on the lead | The vocal can't move closer or further back; the wash is glued to the voice | Send dry. Bounce a separate wet version if you love it, named _Wet
|
| Compression / limiter printed | Flattened consonants, pumping that no plugin undoes | Clip gain and volume edits only; leave dynamics to the mix |
| Peaks at 0 dBFS or "normalized" | Clipped S's and T's, no headroom for the chain | Clip-gain the loudest phrase to about −6 dBFS peak; never normalize stems |
| 16-bit or MP3 files | A noise floor under every plugin stage, pre-echo on transients | 24-bit WAV at the session sample rate |
| Stereo file from a single mic | Double the size, and widening / mid-side plugins react to "stereo" that isn't there | Export mono for mono sources |
| Breaths deleted | Gaps between phrases that sound edited, robotic phrasing | Lower breaths 6–10 dB with clip gain, keep them |
| Comp not finished, alternate takes still active | Two leads fighting; the engineer picks a take you didn't want | Finish the comp, mute or delete alternates before export |
Audio 12_bip.wav |
An hour of soloing to work out which file is the hook double |
Song_Part_v2.wav with the part in words: Lead, Double_L, Harm_High, Adlib |
| Beat at 44.1 kHz, vocals at 48 kHz | Vocal drifts out of time against the beat after a few bars | Export at the session rate and write the rate in your note |
| No rough mix included | The engineer guesses your taste; revision round one is wasted on that | Include the rough you've been listening to plus two sentences of what you like and hate about it |
Related reads & tools
How to prepare vocal stems for mixing (and why each rule exists)
Learning how to prepare vocal stems for mixing is mostly about removing guesses. An engineer opening your folder has to answer four questions before they touch a fader: where does each file start, what was already done to it, how loud is it, and what is it. Every item in the checklist closes one of those questions. Get them right and the first hour of the session is spent mixing. Get them wrong and the first revision round is spent fixing the hand-off instead of the song.
Why dry stems
A reverb or compressor you printed is a decision nobody can undo. Reverb glued to the lead means the vocal can't sit closer or further back than you left it. A printed compressor with the wrong attack eats the consonants the mix needed to cut through the beat. "Dry" means comped, edited, clip-gained and possibly tuned, with no dynamics or time-based processing. If a plugin is the sound — a specific saturation, a telephone EQ on the bridge — bounce that version as well and put it in the name.
Why every file starts at bar 1
Audio files don't carry their timeline position. A file that begins where the first phrase begins will land at the start of the engineer's session, early by however much silence you trimmed. Exporting from a common point, bar 1 in nearly every DAW, makes every file the same length, so the whole folder drags in and lines up without a single nudge. The same applies when you're moving to a different DAW yourself.
Why 24-bit at the session sample rate
24-bit gives about 144 dB of theoretical range, so a vocal peaking at −6 dBFS still has more resolution than any microphone or preamp can use. 16-bit works, but its noise floor is 48 dB higher and every plugin stage adds a little of its own. There's no reason to start lower. Keep the session's sample rate rather than converting on export; mismatched rates between beat and vocals are the classic reason a vocal drifts out of time after a few bars. Peaks around −6 dBFS leave room for the chain; if you're not sure where your files sit, the gain staging calculator gets you there. Don't normalize: it changes the relationship between doubles, harmonies and lead that you set while recording.
Mono or stereo for vocals?
One microphone is a mono source, so export it mono. A stereo file of a mono source doubles the size and, worse, makes some widening and mid/side plugins behave as if there were real stereo information. Stereo is only correct for things that are actually stereo: a stereo pair, or an effect you deliberately printed.
Should I print Autotune?
Yes, if the tuning is the sound: hard-tuned pop, trap, hyperpop. The engineer's tuning will never match your retune speed, humanize and key settings, and that character is part of the record. Send both, Lead_Vocal_Tuned and Lead_Vocal_Dry, so they can use yours and still fix a note you missed. For transparent correction you don't want to hear, send dry, tell the engineer the key, and let them tune as part of the mix, where they can also fix timing.
Naming
Song, part, version. Song_Lead_Vocal_v2.wav. "Vocals final 2 (use this).wav" becomes unreadable the moment a v3 exists. Use the same short song name in every file, number versions, and write the part in words the engineer uses: Lead, Double_L, Double_R, Harm_High, Harm_Low, Adlib. If a file is tuned or has an intended effect printed, that goes in the name too.
SongName_Lead_Vocal_Dry_v2.wav
SongName_Lead_Vocal_Tuned_v2.wav
SongName_Double_L_v2.wav · SongName_Double_R_v2.wav
SongName_Harm_High_v2.wav · SongName_Harm_Low_v2.wav
SongName_Adlibs_v2.wav
SongName_Rough_Mix.wav · Instrumental/ (or beat stems)
NOTES.txt — 140 BPM · F minor · 48 kHz 24-bit · all files from bar 1
What a mixer actually needs so they don't ask you five questions
These are the emails that go out when a hand-off is incomplete. Skip them by including:
- The rough mix with vocals. Your intent is audible instead of described.
- BPM and key in the note, not only in your head. Unsure? Use the BPM detector and song key finder.
- Which file is which: doubles vs harmonies vs adlibs, and whether anything has processing printed.
- One reference record the vocal should sit next to.
- What you like and what you hate about the rough, two sentences each. This saves the whole first revision.
An honest note on the Vocal Chain Bible
The Bible is written for people mixing their own vocals: 88 real chains with the mic, preamp, plugins and settings, each with a free-plugin version. If you're handing stems to an engineer, you don't need it for this song. The free checklist above is the whole value here, and your engineer will choose their own chain. It becomes relevant the day you decide to mix the next one yourself.
FAQ
What format should vocal stems be in for mixing?
24-bit WAV (or AIFF) at the session's sample rate, usually 44.1 or 48 kHz. Mono files for single-mic vocals, every file exported from bar 1 so they're all the same length, peaks around −6 dBFS, no normalizing, no MP3.
Should I send vocals wet or dry to a mixing engineer?
Dry: comped, edited and clip-gained, with no reverb, delay or compression. If an effect is part of the sound, send the dry file plus a separate wet version and say so in the file name and your note.
Should I tune vocals before sending them to be mixed?
If the tuning is the sound (hard-tuned pop, trap), yes, and send the untuned file too. If you want transparent correction, send dry with the key written down and let the engineer tune within the mix.
How loud should vocal stems be?
Peaks between −6 and −3 dBFS on the loudest phrase, set with clip gain so the quiet parts stay naturally quieter. Don't push them to 0, don't limit, and don't normalize; loudness targets like LUFS are for the finished mix, not the stems.
Do I need to send the instrumental as separate stems?
If you have them, yes. The engineer can only balance what is separate, and a vocal fights a 2-track beat differently from one with its own drums and bass. If all you have is a 2-track instrumental, send that, from the same start point as the vocals.
Should I remove breaths from the vocal?
No. Lower them 6–10 dB with clip gain if they're loud; deleting them leaves gaps that sound edited. The engineer can take them further if the genre wants it. If you'd rather leave breath handling entirely to them, say so in the note.
Can I bounce doubles and harmonies together into one file?
Don't. Each double and each harmony part on its own file lets the engineer pan, tune and balance them against the lead. Submixes lock in decisions you made on headphones at 2 a.m.
Mixing them yourself next time?
When you decide to keep the mix in-house, the Vocal Chain Bible is the shortcut: 88 real vocal chains from hit records, as reported by the engineers who made them, with the mic, preamp, plugins and settings, and a free-plugin version of every chain. Your stems are already prepped the way those engineers would want them.
187-page PDF · paid and free-plugin versions · Try one free chain firstFree download
Get the free Grammy Sauce PDF
The plugins and settings 100+ Grammy-winning engineers actually use — collected from 5 years of interviews, in one free PDF.
Want the full pro setup? Get the Vocal Chain Bible — 88 real chains →