Bandry

Production

How to Export Stems for Mixing (So Your Engineer Doesn't Have to Ask Twice)

By the Bandry Team  ·  Sep 2, 2026  ·  6 min read

You booked the engineer, sent a zip file, and three days later you have an email instead of a mix. Files are different lengths. Two of them are named "Audio 12" and "Audio 12_2." The vocal is clipping. There is a limiter still on the master. Now you are re-exporting on a Tuesday night, the turnaround has slipped a week, and the goodwill you had with a busy engineer is spent on housekeeping.

Stem prep is the least glamorous hour of a record and the one that protects everything you paid for. Done properly it takes about forty minutes and the engineer never has to ask you a single question. Done badly it costs days, and on hourly work it costs money.

What a Stem Actually Is

A stem is one element of your song bounced out as its own audio file, aligned to the same start point as every other file in the set. Drag them all into a blank session at zero and the song reassembles itself.

There is a terminology split worth knowing. Strictly speaking, multitracks are every individual channel — each mic on the kit, each guitar layer, each vocal comp — and stems are grouped submixes, like a single stereo file for all drums, one for all guitars, one for all vocals. In practice almost nobody uses the words that carefully, and an engineer who asks for "the stems" nearly always wants the individual tracks so they can actually do the job.

Send grouped stems only when someone specifically asks for them, which usually means a remix, a live backing track, or a mastering situation. For a mix, more separation is better. An engineer cannot un-glue a drum bus, but they can always group tracks themselves.

The One Rule That Breaks Everything Else

Every file starts at bar 1, beat 1, and every file is the same length. Even the tambourine that only appears in the last chorus. Even the vocal that comes in at 0:45. That leading silence is not wasted space; it is the alignment.

Most DAWs will do this for you if you ask correctly. In Pro Tools it is Bounce Mixdown or Commit with the timeline selected from the session start. In Logic it is File, Export, All Tracks as Audio Files with "Include Length of Full Song" or an equivalent selection. In Ableton it is Export Audio/Video with Rendered Track set to All Individual Tracks and the loop brace covering the whole arrangement. In Reaper it is Render, Stems (selected tracks), with the bounds set to Entire Project. In FL Studio it is Export, WAV, with Split Mixer Tracks checked.

The names differ, the principle does not. Select the entire timeline, not the region containing audio, and let the silence print. If you open the exported folder and the file sizes vary wildly, something trimmed itself and you need to go back.

Format, Sample Rate, and Headroom

  • WAV or AIFF, 24-bit or 32-bit float. Never MP3, never AAC, never a lossy format of any kind. If you tracked at 32-bit float and your DAW offers it, use it; there is no downside beyond file size, and it makes accidental clipping recoverable.
  • Same sample rate as the session. If you recorded at 44.1 kHz, export at 44.1 kHz. Converting up invents nothing and doubles the transfer. Converting down discards information the engineer may want. Do not "helpfully" change it.
  • Take the master bus processing off. The limiter, the bus compressor, the mastering-chain plugin you added to make the rough feel finished — all of it comes off before you export individual files. Otherwise the same processing prints onto every stem and gets applied a second time later.
  • Leave headroom. Your loudest single file should peak somewhere around -6 dBFS. Nothing should touch 0. This is not superstition; the engineer is going to sum thirty files together and needs room to do it.
  • Mono stays mono. Do not export a mono kick drum as a stereo file with two identical channels. It doubles the file size and makes panning decisions slightly more annoying downstream. Stereo sources — overheads, a wide synth pad, a stereo room mic pair — stay stereo.

Naming Them So Nobody Has to Guess

Naming is where the actual respect for the engineer's time shows up. They are going to open a folder and start dragging. Every file that makes them stop and listen to figure out what it is costs a minute, and there are forty files.

Use instrument, then position or mic, then take number if it matters. No spaces, no slashes, no emoji, nothing that will trip up a different operating system:

  • Drums: KickIn, KickOut, SnareTop, SnareBottom, HiHat, Tom1, Tom2, OHL, OHR, RoomMono
  • Bass: BassDI, BassAmp
  • Guitars: GtrRhythmL, GtrRhythmR, GtrLead, GtrClean_Verse
  • Keys and synths: PianoL, PianoR, SynthPad, OrganMono
  • Vocals: VoxLead, VoxLeadDouble, VoxHarmHigh, VoxHarmLow, VoxAdlib1

Put the whole set in one folder named with the song title, the tempo, and the key: SongTitle_128bpm_Gmin. If you are sending more than one song, one folder per song, never a single flat pile.

Dry and wet: for anything with a creative effect that is genuinely part of the part — a specific amp sim, a delay that plays a rhythmic role, a pitched vocal effect — print it twice. VoxLead and VoxLead_FX. The engineer will nearly always use your wet version, but if it fights the mix they have a way out that does not involve emailing you.

What Goes in the Folder Besides Audio

The files are half the delivery. The context is the other half, and it is what separates a session that goes smoothly from one that runs on guesswork.

  • Your rough mix, as a single stereo file, clearly labelled ROUGH. This is the most useful thing in the folder. It tells the engineer your intended balance, which parts you think are important, and roughly what the song is supposed to feel like. They will move away from it, but they need to know where you started.
  • Tempo and key, plus a note if the tempo moves. If there is a tempo map or click, include it.
  • One or two reference tracks, commercially released songs whose sound you are aiming at. Be specific about what you want from each one — "the vocal sits like this," "the low end is this tight" — rather than just naming a band.
  • A short notes file. Which effects are intentional, which parts you are unsure about, anything you specifically do not want (too bright, too compressed, too much reverb on the lead). A paragraph is plenty. Strong opinions are more helpful to an engineer than none.
  • A cleanly labelled transfer link. Zip the folder, send one link, and make sure the link does not expire in seven days. Nothing derails a session like a dead download on day nine.

The Pre-Send Checklist

Before you send anything, open a fresh session and import your own export. Five minutes here catches almost every problem that would otherwise become an email.

  • Drop every file in at zero. Does the song play back correctly with nothing moved?
  • Are all files identical in length?
  • Does any file clip, and is the loudest one peaking somewhere near -6 dBFS?
  • Is any file silent all the way through, meaning a muted track printed empty?
  • Can you tell what every file is from its name alone, without listening?
  • Is the rough mix in the folder, and is the master-bus limiter off the individual stems?

If all six pass, you are done. If your song still needs structural work or the arrangement is not settled, stop — that is a producer conversation, not a mix one, and no amount of clean stem prep fixes it. And if you are mixing it yourself instead of hiring out, the order of operations for mixing a song is the next thing to read.

Finding Someone to Send Them To

Mixing is remote work, so on Bandry mixing and mastering engineers post as remote and their listings are visible everywhere, not just in your city. Engineers post their own listings under Offering, with links straight to their released work, so you can hear what they have actually done before you say a word to them. If you would rather describe the project and let people come to you, post it yourself under Seeking.

When a listing looks right, you tap 🔗 to link up, which hands your contact details to that engineer directly. There is no messaging layer and no platform sitting between you and the booking. You email them, they quote you, and you send the folder you just built. The find-a-mixing-engineer page walks through the whole flow, and when the mix comes back, the mastering step is the last hop before release.

Frequently asked questions

What are stems in music?

In everyday practice, a stem is one track from your session bounced out as its own audio file, starting at the exact same point as every other file. Purists distinguish stems (grouped submixes like 'all drums') from multitracks (every individual channel), but when a mixing engineer asks for your stems they almost always mean the individual tracks. If you are unsure which they want, ask — it takes one message and saves a re-export.

Should stems start at the beginning of the session or at the first note?

Always at the very start of the timeline, bar 1 beat 1, even if the track is silent for the first 30 seconds. Every file must be exactly the same length. That way the engineer drags all of them in at zero and the whole song lines up automatically. Trimming files to the first note is the single most common mistake and it forces someone to re-align everything by ear.

What sample rate and bit depth should I export stems at?

Export at the same sample rate the session was recorded at — do not convert. Use 24-bit or 32-bit float WAV or AIFF. Never send MP3s or other lossy files for mixing. If your session was tracked at 48 kHz, send 48 kHz files; converting up to 96 kHz adds nothing but file size, and converting down throws away resolution the engineer might want.

Should I remove plugins and effects before exporting stems?

Print two versions of anything with a creative effect: one dry and one with the effect baked in. Remove any compressor, EQ, or limiter you put on the master bus, and remove corrective processing you are not committed to. Keep effects that are part of the sound itself, such as a specific guitar amp sim or a delay that plays a rhythmic role, and tell the engineer which ones are intentional.

How much headroom should stems have?

Aim for peaks around -6 dBFS on your loudest individual file and make sure nothing clips. Take the limiter off the master bus entirely. With 24-bit or 32-bit float files there is no noise-floor penalty for leaving headroom, and the engineer needs room to work. A stem that already peaks at 0 dBFS gives them nowhere to go.

How should I name stem files?

Use a consistent pattern: instrument, then position or mic, then take number if relevant. KickIn, KickOut, SnareTop, SnareBottom, BassDI, BassAmp, GtrRhythmL, GtrRhythmR, VoxLead, VoxHarm1. Avoid spaces and special characters, avoid DAW defaults like 'Audio 37', and put the whole set in one folder named with the song title, tempo, and key.

Ready to try it?

Bandry is live on the App Store. Download it and start posting.

Download

Follow Bandry