Stems vs Multitracks: Prepare an Audio Handoff

What is the difference between stems and multitracks?

Stems combine related tracks into grouped audio files; multitracks keep individual elements separate for further mixing. Before export, agree on the required control, processing, shared start and ending, file format, and reference mix. Work from a backed-up session and share only material the recipient is permitted to receive. Loud monitoring can cause permanent hearing loss: reduce sound level and exposure time, and use appropriate hearing protection at 85 dB(A) or above.

The delivery question is not "How many files can this project produce?" It is "Which decisions should the recipient be able to change?"

Which package gives the recipient the right control?

iZotope's terminology guide distinguishes mixed groups from the individual elements used to make a mix. Use the distinction, but confirm the recipient's meaning in writing; filenames alone do not describe processing.

A track count is not a quality measure. Too little separation prevents the requested work; unnecessary alternatives make the approved choices harder to identify.

Ask the recipient to name the parts they expect. "Vocals" might mean a combined vocal group, or separate lead, doubles and backing vocals. Resolve that before committing the files.

For a single stereo mix rather than a separated package, use the mastering-export guide.

What should the delivery agreement specify?

Write a short export brief before changing the session:

• Scope: grouped stems, individual tracks, or both in separate folders.
• Processing: exactly which edits, tuning, automation, inserts and shared effects are included.
• Timing: one common start, one complete ending, and any tempo-map information needed.
• Format: recipient-approved file type, sample rate, bit depth and channel layout.
• Reference: the approved mix version and any intentional difference from it.
• Rights: which recordings, rendered instruments and supporting files may be shared.

Preserve the original session. Save a delivery version so that removing a processor for a requested alternate does not overwrite the approved mix.

Do not treat "dry" as a complete specification. It may mean no reverb, no shared effects, or no processing at all. Replace the label with a sentence such as "Edited lead vocal, tuning retained, no reverb or delay, no Main processing."

Our project-file organization guide covers the archive and backup structure around this agreement.

How should common start and end points work?

Export every contribution across the agreed complete interval, including silence before a late entrance. Do not trim each file to its first audible event unless a different alignment method has been explicitly agreed.

Ableton's stem-export guidance specifies matching Render Start and Render Length across the arrangement regardless of where an individual part begins or ends. This preserves an alignment reference when files move between projects.

Use enough ending time for the actual agreed decay. There is no universal tail length.

For an original hypothetical example, suppose the musical section occupies 96 seconds from the agreed start and the chosen ending includes another 4 seconds of decay. Every delivery contribution spans 100 seconds.

At an agreed 48,000 sample frames per second:

100 seconds x 48,000 frames per second = 4,800,000 sample frames

A stereo file contains two channel samples per frame. The frame count still describes 100 seconds; do not double the duration because there are two channels. These are checked arithmetic assumptions, not measurements from a rendered session.

What would a clear stem manifest look like?

The following is an invented handoff plan. No files were rendered or auditioned for this example. It assumes the recipient requested five musical groups plus a separate shared-effects return.

Each file has the same 100-second interval. The manifest also records the agreed sample rate, bit depth, channel format and exact session version; these are delivery decisions, not universal defaults.

Include the approved reference mix separately. That makes six contributions and one reference, not seven contributions to be played together. Label the reference folder accordingly.

If the recipient instead needs individual drum control, this manifest is the wrong package. Replace the grouped drum file with the agreed individual tracks and update the inventory; do not silently add them beside the drum stem as simultaneous inputs.

Where should shared reverb and delay go?

Choose one documented arrangement. Effects may be included with their source group, supplied separately, or provided as clearly labeled alternatives according to the recipient's request.

The manifest uses a separate shared return. Therefore, its direct groups exclude that return. If an export already includes the same return contribution and the recipient adds it again from another file, that contribution is duplicated.

This is a bookkeeping check, not a claim that every DAW handles sends identically. Inspect the actual render source and routing rather than inferring them from the track name.

Do not include both a parent group render and its child tracks in the same reconstruction unless the plan intentionally calls for that. Ableton's help page notes that its individual-track export can include grouped tracks at group level with group processing, plus the individual children without that group processing. The recipient needs to know which set constitutes the intended mix.

What is specific to Ableton Live 12?

The Live 12 export manual documents All Individual Tracks as post-fader outputs, including returns and MIDI tracks with instruments. It is not a command that automatically produces unprocessed recordings.

The same manual describes Include Return and Main Effects as rendering each selected track with its used returns and Main effects. Check this option against the delivery brief. Do not enable it automatically merely because the files are called stems.

Normalize changes the rendered file's gain according to its peak. Leave normalization off when the agreed package is meant to retain its existing relative levels, as Ableton's stem-export instructions specify.

These names and behaviors are scoped to the cited Live 12 documentation. Other programs and versions need their own export checks. No interface test is claimed here.

Must the stems reconstruct the reference mix exactly?

Set that expectation before export. A sound-preserving stem package and a flexible remix package need not make the same compromises.

iZotope notes that nonlinear processing on the full mix, including compression, saturation and limiting, can prevent separately rendered stems from adding up exactly to the original. The full combination and isolated contributions do not necessarily receive equivalent processing.

Do not remove a creative mix-bus chain solely to make a comparison easier. Discuss the intended difference with the recipient and preserve the approved reference. If exact reconstruction is required, have the responsible engineer verify the routing and rendering method.

For instrument parts, distinguish rendered sound from editable performance instructions. The MIDI guide explains why a MIDI file alone does not preserve the chosen instrument sound.

How should the receiving check work?

Use a separate verification project. Import only the intended contributions at the common origin. Confirm file count, duration, sample rate, channel layout, leading silence and ending before evaluating the musical balance.

Check the receiving application's automatic time-stretching and import behavior through its documentation. Do not assume that dropping equal-length files into a project preserves every intended setting.

Compare against the reference as an alternative playback source, not an additional track in the sum. Check entrances, edits, effects and the final decay. Record differences; do not fix them by making undocumented gain changes.

Keep monitoring controlled throughout. NIOSH identifies 85 dB(A) as an eight-hour occupational time-weighted exposure limit, with allowable time halved for each 3 dB increase. It is not a recreational-listening safety guarantee. Reduce level and exposure time; use appropriate hearing protection at 85 dB(A) or above. Do not increase playback level to inspect quiet tails.

Return a concise acceptance note: received file list, alignment confirmed or unresolved, reference version, and specific questions. Retain the backed-up source session and the exact reviewed exports. Follow equipment manuals, and share only files the recipient's rights or permissions allow.

Sources

• iZotope: Stems and multitracks - separation, processing and reconstruction limits.
• Ableton: Importing and exporting stems - shared interval and grouped/individual export checks.
• Ableton Live 12: Managing Files and Sets - render-source and effect-inclusion behavior.
• NIOSH: Understanding Noise Exposure Limits - hearing-risk controls and occupational-limit scope.