Tutorials Audio

Stems vs Multitracks: What They Are and When to Deliver Each

Beginner · ~20 min

Overview

"Can you send me the stems?" and "Can you send me the multitracks?" sound similar but mean different things, mixing them up leads to sending the wrong deliverable and a frustrating back-and-forth. This guide clears up the terminology and gives you a repeatable export workflow for both.

What You Need

Your finished project session in your DAW
Clarity on who's requesting the files and why (mix engineer, video editor, remixer, archive)

Steps

Understand what a multitrack actually is

A multitrack export is every individual raw recorded track (each mic, each speaker, each instrument) completely unprocessed and untouched, exactly as it was recorded, all starting at the same timeline position.

Understand what a stem actually is

A stem is a submix, several related tracks (all drums, all backing vocals, the full musical bed) bounced down into one processed, mixed file. Stems already have your EQ, compression, and levels baked in. Multitracks don't.

Know who asks for which

A mastering engineer or remixer usually wants stems (a manageable, already-mixed set of groups). A video editor syncing dialogue to picture, or another engineer doing a from-scratch remix, usually wants multitracks (full raw access to every source).

Bounce multitracks correctly

Export each track individually with no processing (or only essential corrective processing, clearly noted), all starting at sample 0 or a documented common start point, and named clearly (Host_Lav, Guest_Lav, Room_Mic).

Bounce stems correctly

Group related tracks into logical submixes (Dialogue, Music, SFX for video, Drums, Bass, Vocals, Music for a song), print each stem with your final mix processing applied, and keep every stem at the same length and start point as the others.

Pro Tips

Always keep every exported file the same length as your full mix, even if a track is silent for most of it, mismatched lengths break sync when someone else imports your files.
Deliver a rough reference mix alongside stems or multitracks so the receiving engineer has a sonic target to match.
Name files with no spaces or special characters, cross-platform delivery (Mac to Windows, DAW to DAW) handles simple underscored names far more reliably.

Knowledge Base

The Core Distinction

Multitracks = raw, individual, unprocessed. Stems = grouped, processed, submixed. If someone can't tell you which one they need, ask what they're doing with the files, remixing or re-mixing from scratch needs multitracks. Adding to an existing mix (a video edit, a mastering pass) usually needs stems.

Common Stem Groupings

For music: Drums, Bass, Instruments, Lead Vocal, Backing Vocals.
For video/podcast: Dialogue, Music, Sound Effects/Ambience.

Why Getting This Wrong Wastes Everyone's Time

Sending stems when multitracks were needed means the receiving engineer can't isolate or remix individual elements. Your processing is permanently baked in. Sending multitracks when stems were needed dumps unmixed raw chaos on someone expecting a near-final sound. Confirming which one before exporting saves a round trip.

Where This Fits

This guide covers one specific part of mixing and mastering. The wider picture. The full mix workflow from gain staging and balance through panning, bus processing, and knowing when to stop, is in Mixing a Full Song: From Rough Levels to Final Bounce, which frames the discipline as a whole and links out to the detailed guides underneath it, including this one. If you are starting from scratch rather than solving a specific problem, read that first and come back here.

FAQ

Q: Should stems include any processing at all, or just level?
A: Include your actual mix processing (EQ, compression, level). A stem is meant to represent a finished, mixed group, not a raw group at a rough level.

Q: Is there a standard sample rate/bit depth for delivery?
A: Match your project's native sample rate and deliver at 24-bit WAV unless the recipient specifies otherwise. This preserves headroom for further processing without unnecessary bit-depth reduction.