Music Stems: Agree What the Handover Contains

A filmmaker asks a musician for stems so the music can sit beneath a spoken introduction. The musician sends every individual instrument track. The editor expected a handful of balanced groups and now has a mixing job to do before cutting the film. The files may be perfectly usable, but they answer a different brief.

For independent musicians, video creators and collaborators, the handover should specify what the recipient needs to change. A folder labelled stems leaves several decisions unresolved: which sounds belong together, which effects are included, where the files begin and what reference they should match. Resolve those decisions before the export.

Grouped stems and individual tracks serve different tasks

In iZotope’s guide to stems and multitracks, dated 19 September 2024, Nick Messitte describes a stem as related tracks mixed together into one file. His example contrasts individual drum recordings with a rendered drum submix. Multitracks provide the separate elements used for mixing.

The practical question is how much control the recipient needs. A grouped drum file provides one delivered component for the drum balance. A set of individual drum tracks supplies different components for a more detailed mix. Neither handover is automatically better; choose the one that supports the agreed work.

Rather than settling a terminology argument, write the expected components into the brief. If a collaborator uses the word stems to mean individual tracks, an explicit file list can expose that difference before either person spends time preparing the wrong package.

Start with the intended edit

Consider a fictional 60-second film score made for an artist’s studio video. The editor wants to lower the lead vocal beneath an introduction, retain the instrumental balance and make the drums quieter for the closing shot. This example proposes a workflow; no audio project or export was tested for the article.

For that brief, the collaborators agree on four grouped files: drums, bass, other instruments and lead vocal. They also request a full reference mix. The editor expects to adjust those groups, rather than rebalance every microphone or synthesiser part.

Now change the request. Suppose the editor specifically wants a quieter snare while keeping the kick drum unchanged. The proposed four-file package does not supply those two instruments as independently adjustable files. Ask the musician whether a revised drum stem or additional individual tracks would serve the edit. Do not promise a control that the delivery list does not provide.

An assistant could turn the agreed requests into a checklist, but the musician and editor should approve the meaning of each group. This is a suggested administrative use of AI, not a claim that a particular assistant can identify every sound in a mix or produce suitable exports.

Name the processing included in each file

Export settings can change the delivered material. The Ableton Live 12 reference manual describes an Include Return and Main Effects option for individual or selected track exports. Enabling it includes the relevant return signals and Main-track effects. The manual also distinguishes individual-track outputs from the Main output.

For the fictional score, the lead vocal has a prominent echo. Agree whether that echo is part of the vocal delivery, provided separately or deliberately omitted. A note that says vocal with the approved echo is more specific than a filename saying final vocal.

Make the same decision for the other groups. The purpose is to let the recipient understand what they are receiving, not to impose one universal setting. A film editor preserving an approved sound and a mixer rebuilding the production may require different packages.

Include the application version and a concise account of the settings actually used when the export is performed. The documented Ableton option establishes a capability of that application; it does not establish how a file from another application, or an unchecked session, was rendered.

Agree a shared timeline and ending

In the fictional score, the vocal first enters 12 seconds after the beginning. The agreed package keeps every file on the same timeline, including the opening silence before that vocal. If someone trims the vocal file to its first sound and imports it at the start, the planned 12-second entry is no longer preserved by that placement.

Ableton’s manual provides Render Start and Render Length controls and says its All Individual Tracks exports have the same length. For a handover, record the intended range and inspect the actual files. A named export option is useful context for that check, rather than a substitute for it.

The film’s picture ends at 60 seconds, but the collaborators want to preserve an echo through 64 seconds for the editor to review. They therefore agree a common export range ending at 64 seconds. Those timings are invented requirements, not recommendations about every score or claims about an audio file produced here.

Check the package as the recipient will receive it

For this proposed project, the musician should keep the original session and review a separate copy of the delivery folder. The following are suggested acceptance checks, with the agreed brief beside them.

• Confirm that the four agreed groups and the reference mix are present and clearly named.
• Check the format, sample rate, bit depth and channel arrangement against the recipient’s request.
• Import the delivered groups at the agreed starting point in a test session.
• Listen through the introduction, the vocal entry and the ending, including the requested tail.
• Try the actual requested edits and note whether the supplied groups provide the necessary control.

Compare the combined groups with the reference mix under the stated playback setup. If they differ, investigate the routing and export choices before describing them as a match. The reference is a review target, not evidence that any arbitrary collection of separately rendered files reproduces it.

Keep the version note short and concrete: which score revision was exported, what each group includes, the shared start and end, and which checks were completed. Distinguish an instruction to perform a check from a recorded result.

Deliver a usable creative relationship

File contents are one part of a music handover. MaryChuks.com’s guide to a release rights pack addresses separate questions about permissions, contributors and metadata. A clear stem list does not settle those matters, and a complete rights record does not establish that the editor received the audio components requested.

The useful promise is a package that supports named changes. Agree the groups, processing and timeline, then check the exported folder against that agreement. That gives both collaborators a shared basis for the next edit and makes a request for stems specific enough to fulfil.

Sources

Primary guidance checked on 31 October 2026. The score, film, file list and timings are hypothetical. No audio export, mixing application or AI audio service was tested.

• iZotope: Stems and multitracks: What’s the difference? — Nick Messitte, 19 September 2024.
• Ableton Live 12 Reference Manual: Managing Files and Sets, exporting audio and video — maintained documentation; no individual publication date asserted.