How to Send Stems to a Mixing Engineer

To send stems to a mixing engineer, export clearly named lossless audio files over one shared timeline range, keep the original sample rate, and include a reference mix plus a short technical brief. Confirm the engineer's preferred bit depth and processing choices before exporting.

Ableton's current stem export guidance gives the DAW-neutral core: render the full arrangement, keep every file the same length, avoid normalization, use PCM WAV or AIFF, and match the source sample rate.

One terminology check prevents confusion: some engineers use stems for grouped submixes such as drums, guitars, and vocals, and multitracks or parts for each individual track. Ask which deliverable they want before you print dozens of files.

File Names That Don't Make Anyone Cry

Engineers open a folder, sort by name, and start importing. If your file names look like Audio 1.wav, Audio 1_2.wav, bv copy FINAL_v3.wav, the first thing they do is rename everything. That's twenty minutes of your money, and a real chance they put a vocal where a hi-hat should be.

A naming convention that actually works:

01_Kick.wav
02_Snare_Top.wav
03_Snare_Bot.wav
04_HatClosed.wav
05_OH_L.wav
06_OH_R.wav
10_Bass_DI.wav
11_Bass_Amp.wav
20_GuitarRhythm_L.wav
21_GuitarRhythm_R.wav
30_LeadVox.wav
31_LeadVox_Double.wav
40_BackingVox_Stack.wav

Three rules. Number them in a logical order (drums first, vocals last, lead before backing). Use underscores, not spaces. Don't use slashes, brackets, or accented characters, some systems still mangle them.

If you have multiple takes or alternates, suffix the file (_takeA, _takeB) and tell the engineer which one is the keeper. Don't make them guess.

Format: Lossless PCM, Agreed Bit Depth, Same Sample Rate

Use lossless PCM WAV or AIFF unless the engineer requests another format. Do not use MP3 or AAC as the mix source because those formats discard audio data.

Bit depth: ask the engineer. 24-bit PCM is common, while some workflows prefer 32-bit float to preserve the DAW's render precision. Do not add dither unless the engineer requests it.

Sample rate: whatever the session was recorded at. If you tracked at 48kHz, export at 48kHz. Don't sample-rate-convert on the way out, you'll bake in a quality hit for no reason. Tell the engineer the sample rate in your message so they set up the session to match.

Stems Should Start at Bar 1, All of Them

This is the rule that saves the most time. Every file starts at the same point, even if the instrument doesn't come in until the second chorus. That way the engineer drags every file to bar 1 in the session and they all line up automatically.

If you trim each file to just the bits where the instrument plays, you've made yourself look efficient and made the engineer's job much harder. Don't trim. Export the full song length for every track.

Most DAWs call this "consolidate from bar 1" or "export from session start." Use it.

Headroom: Leave Some

Do not normalize the exports and do not let them clip. There is no universal peak target for every source; preserve the session's healthy gain structure and confirm any delivery target with the engineer.

If your mixbus is hitting a limiter on the way out, bypass the limiter for stem export. Same for any glue compression on the master bus that you don't want printed in. The whole point of sending stems is that the engineer is doing the mixing, not you. Hand them the raw ingredients.

Stem effects are a separate question, see the next section.

What to Print With Effects, What to Print Dry

The honest answer is: ask the engineer. But the default that works for most modern productions:

Print with effects:

• Vocal tuning (Melodyne, Auto-Tune), bake it in, you don't want them re-tuning
• Sound design FX, risers, impacts, designed transitions
• Anything where the effect is the part (filtered bass that's the whole hook, formant-shifted backing vox)

Print dry, send the wet version too:

• Vocal reverbs and delays, give them dry and a separate "FX print" file. The engineer will probably re-do reverb but might want to hear what you intended
• Guitar amps, if you tracked DI, send DI and amp print
• Drum room mics, keep room mics as their own track, don't bake them into close mics

Usually leave off unless requested:

• Mixbus limiting or processing that would be applied repeatedly to every individual export
• Master-fader processing that the engineer needs to rebuild at mix stage

If in doubt, send both versions and label them clearly: 30_LeadVox_Dry.wav, 30_LeadVox_FX.wav.

The Reference Mix Is Not Optional

Send a stereo bounce of your demo mix alongside the stems. This is the single most useful file in the whole package. The engineer wants to know what you heard in your head, what balance, what energy, where the chorus lifts.

Two versions help even more:

• _RefMix.wav, your rough mix as it stands
• _RefMix_LoFi.mp3, same thing as a 320 MP3 for quick phone listens

Add a short note: "vocal sits a bit forward, drums could be punchier, the reference vibe is somewhere between Phoebe Bridgers and Big Thief." Reference tracks from other artists help too. Don't be precious about saying "I want it to sound like X", engineers love that brief.

Metadata: BPM, Key, Tempo Map

Every session needs these three numbers. Put them in the message and in a tiny _README.txt inside the folder:

Track: Long Way Home
BPM: 92
Key: F# minor
Sample rate: 48kHz / 24-bit
Notes:
- Tempo is constant (no map needed)
- Vocal comp is a single take, no edits done
- The "ooh" stack at 2:14 is two voices, both me
- Open to ideas on the bridge, feels long to us

If the song has tempo changes, export a tempo map (most DAWs export MIDI tempo maps) and include it. If there's a click track, include the click as its own file in case they need to re-grid anything.

Key matters because pitch correction and tuning plugins want it. BPM matters for any time-based effect the engineer adds. They will work it out from the audio if they have to, but you knowing this off the top of your head buys you respect.

What Not to Send

• The full DAW project file (unless they explicitly asked, most engineers don't want to open your Logic session in their Pro Tools rig)
• Hundreds of unused alternate takes, ten reverb returns nobody used, the ghost of a kick drum you muted weeks ago. Clean the session before exporting
• A WeTransfer link that expires in seven days, right before they get to it
• Twelve separate emails, each with two stems attached

One folder, all stems, reference mix, README, sent in one link. Done.

The One-Link Way

The reason most stem deliveries go sideways isn't the stems, it's the delivery layer. Engineer downloads files, link expires, you re-upload, comments come back over WhatsApp, the v3 mix gets confused with the v2 mix, and three weeks later nobody remembers what was approved.

Vandall is built for this exact loop. One project link holds the stems, the reference mix, the metadata, and the conversation. Comments anchor to the second on the audio, "the vocal pops at 1:47" is a click, not a description. Versions stack so the v1 mix and the v3 mix both stay reachable. The link doesn't expire.

Drop your project into Vandall and send one link to your engineer. Or read the full feature breakdown if you want to see what's under the hood.

The engineer's job is to mix your record. Yours is to hand them clean files and a clear brief. Do that and the rest is just music.