Beta opens 15 Oct Join
Blog

Sync11 min read

Stems, alts and cutdowns: what libraries and supervisors want

Stems vs alt mixes vs cutdowns explained: what music libraries and supervisors ask for, when to make them, formats, and a music file naming convention.

By Daniel Angelucci, composer and founder of DADAAUDIO

Stems are the separate layers of your mix (drums, bass, strings, vocals), each bounced as its own file at the full length of the track. Alts are complete alternative mixes of the same track (instrumental, no lead melody, drums and bass only), and cutdowns are shorter edits of it (60, 30 and 15 seconds, stings, loops). Libraries and supervisors ask for them for one reason: an editor cutting to picture needs to make your music fit, and these files let them do it without replacing you with another track.

The confusion is understandable, because the words overlap and every library uses them a little differently. "Splits" belongs to a different conversation altogether. Here is how the pieces fit together, what to prepare and when, and how to name the files so nobody has to guess.

Stems vs alt mixes vs cutdowns

Stems

A stem is a group of related parts, submixed and printed as one file. Sound On Sound's guide to library delivery describes them as roughly five to ten stereo files grouping your major sound types such as brass, guitars, bass, drums and percussion. They are not your raw multitrack, and they are not a finished mix on their own.

The one rule nearly everyone agrees on: every stem starts at the same point and is exactly as long as the main mix, with silence wherever that instrument isn't playing. Cinephonix's composer instructions say stems must be "exactly the same length as your main track", and that their system rejects them otherwise. Drop the stems into a session at zero and they should play back as your mix.

Alts (alternative mixes)

An alt is a finished, full-length mix with something taken away or changed. Common ones:

  • Instrumental: the whole track without vocals. For songs, this is the alt you should never be without. Orphiq's sync guide notes that many placements use the instrumental, entirely or wherever dialogue needs to be heard.
  • No lead / underscore / bed: the melody or lead line removed so the track can sit under dialogue. Sound On Sound calls underscore music designed to set a mood without distracting from what's being said.
  • Rhythm, drums and bass, sparse, no drums: ALIBI, for example, offers rhythm mixes (leads stripped out), sparse mixes (no drums), and drums and bass mixes, among others.
  • Clean: lyrics edited for broadcast. If a song has explicit words, have one ready.
  • A cappella: vocals only, mostly for trailers and remixes.

Cutdowns, edits, stings and loops

Cutdowns (also "edits") are shorter versions of the track that still sound like complete pieces: a beginning, the identity of the track, and a proper ending. The standard lengths are 60, 30 and 15 seconds, the lengths of broadcast spots. Sound On Sound asks for "very accurately timed" cutdowns at those three lengths.

How accurate is accurate? On a long-running MusicLibraryReport thread on 15, 30 and 60 second edits, composers disagree on the ring-out, but agree that the last hit should land just under the length (around 14.5, 29.5 or 59.5 seconds) and that going over is the mistake to avoid. If a library gives exact timings, follow theirs.

  • Sting (stinger, button): a few seconds of the track's character, a hit or a chord, used as an ending or to mark a cut. Media Music Now describes stingers and bumpers as short clips that introduce, end or link sections, usually under 15 seconds.
  • Bumper: similar, often used to go in and out of a break.
  • Loop: a section that repeats seamlessly, for menus, games, websites and anything with an unknown length.

And splits?

Splits are not audio at all. They are the percentages of ownership in the song (and separately in the recording) that you agree with your co-writers and publishers. Collecting societies want them registered and adding up to 100 %; STIM, for example, asks you to enter each person's percentage "so that the total amounts to 100 percent". PRS, ASCAP, BMI, SACEM, GEMA, SIAE and SOCAN all work with shares in their own way. A stem has no split; the work and the recording do.

Why editors want them

Picture comes first. An editor has a 47-second scene with a line of dialogue at 0:12 and a door slam at 0:31. Your full mix almost fits. With an underscore mix they keep the dialogue clear; with stems they can drop the drums for the first section and bring them in on the slam; with a 30 they have a natural ending to work from.

ALIBI puts the editor's view plainly: alt mixes and stems "can also potentially save you from having to find an entirely new track". That is the whole case. Every file you deliver is one more reason the track stays in the cut instead of being swapped at the last minute.

What libraries and supervisors typically ask for

There is no universal spec, and libraries differ on things you might expect to be standard. Sound On Sound's guide asks for mixes "unmastered, with no bus compression or bus EQ"; Cinephonix asks for the main track as the "fully mastered best version" and makes the sting and 15/30/60 edits itself from your stems. So always read the delivery sheet. As a working baseline:

What Production music library Supervisor pitch (songs)
Main mix Required, full length Required
Instrumental Usually required Expected; have it ready
Underscore / no lead Often Sometimes
Other alts (no drums, rhythm, sparse) Often, library's own list Rarely at pitch stage
Clean version If lyrics need it If lyrics need it
60 / 30 / 15 cutdowns Often (some libraries make their own) Rarely; editors usually cut
Stings, loops Often Rarely
Stems Often, 5–10 groups On request, often fast

Two practical points. First, some composers on MusicLibraryReport report that larger libraries and ad agencies cut their own 30s and 60s from the full mix, so ask before spending a day on edits. Second, each library wants its own set of alts with its own names, so it usually makes sense to pitch with the main mix and instrumental and make the rest once a track is accepted.

When to prepare them

Prepare while the session is still open, not when the email arrives. Bouncing stems from a session you closed six months ago, with plug-ins that have since updated, is how you end up with stems that don't add up to the mix.

A practical rhythm:

  1. When you finish the mix: print the main mix, the instrumental and your stem groups from the same session, same start point, same length.
  2. When you pitch: send the main mix (and the instrumental for songs). Say stems and alts are available.
  3. When a library accepts the track: read their sheet and make the alts and cutdowns they actually list, named their way.
  4. When a supervisor asks: you already have the stems. Delivering in an hour rather than three days matters on a deadline.

Formats

For picture, the usual request is WAV or AIFF at 48 kHz, 24-bit. Sound On Sound gives exactly that as the delivery format for film and TV use. Some libraries also accept 44.1 kHz, and a few will want MP3 previews for pitching. Again: the delivery sheet wins.

Check before you send:

  • Every file starts at exactly the same point; no silence trimmed from stems.
  • No clicks at the start or end, no cut-off reverb tails.
  • Stems summed together sound like the main mix (no missing master-bus effects, nothing out of phase).
  • Nothing clips.

A file-naming standard that works

A file name is the last piece of metadata that never gets lost. Embedded tags get stripped, spreadsheets get separated from folders, but the name travels with the audio. A good name tells a stranger which track it is, which version, and (for edits) how long, without opening it.

Pattern:

Title_Version.wav for alts and edits, and Title_STEM_Group.wav for stems, with an optional artist or catalogue code at the front if the files will be mixed with other people's.

File Name
Main mix Northern-Line_Main.wav
Instrumental Northern-Line_Instrumental.wav
Underscore Northern-Line_Underscore.wav
Clean Northern-Line_Clean.wav
60-second edit Northern-Line_60.wav
30-second edit Northern-Line_30.wav
15-second edit Northern-Line_15.wav
Sting Northern-Line_Sting.wav
Loop Northern-Line_Loop.wav
Drums stem Northern-Line_STEM_Drums.wav
Strings stem Northern-Line_STEM_Strings.wav

Rules that keep it clean:

  • Title first, so every file of a track sorts together.
  • One vocabulary, the same words for every track you deliver. If the library has its own labels, use theirs. ALIBI, for instance, deliberately avoids labels like "no drums" in its own catalogue for search reasons, while other libraries use them.
  • No "final", "v3", dates or your DAW's suffixes. If you need revisions, keep them in your own system, not in the delivered names.
  • Hyphens inside the title, underscores between parts. Avoid characters that break in some systems: / \ : * ? " < > | and accented letters.
  • Number stems (_STEM_01_Drums) if the library wants them in a fixed order.

Libraries group these files on their side too: SourceAudio, for example, lets a library nest alternates, cutdowns and stems under the main version. Consistent names make that matching quick instead of a guessing game.

Embedded metadata: yes, WAV can carry it (but don't rely on it)

You will read that "WAV files can't hold metadata". That isn't true. A WAV file is built from chunks, and the RIFF format includes an INFO chunk for title, artist, copyright and similar fields. The Broadcast Wave Format (EBU Tech 3285) adds a bext chunk with a description of up to 256 characters, the originator, a reference, date, time reference and loudness values; it is what most professional audio software writes. iXML and even ID3 tags can also be stored in their own chunks.

The real problem is support. Programs are required to ignore chunks they don't understand, so metadata written by one tool may simply not be read by the next. DISCO, used by many publishers and supervisors, says it plainly: ID3v2 writes reliably to MP3 and AIFF, but WAV files "do not reliably carry ID3v2 metadata".

So:

  • Embed the basics anyway (title, composer, publisher, your contact), because when it survives it helps.
  • Always deliver a metadata sheet with the files: title, version, length, writers with IPI and society, publishers, splits, ISRC, BPM, key, and a short description.
  • Make the file name carry the essentials, because the name always survives.

ISRCs for alts and edits

Each version you release is its own recording. The ISRC Handbook says different mixes require different ISRCs, and so do edited versions (a clean edit, for example), remixes and extended versions. An edit that removes verses or choruses gets a new code; a change to the fade alone that alters the length by less than 10 seconds doesn't. Stems made as a normal part of production don't need an ISRC, but a stem or a vocal-less version intended for release does. Many libraries assign their own codes, so ask before you register.

Common mistakes

  • Stems that don't line up: trimmed silence, different start points, or a bar missing at the end.
  • Stems that don't add up: the master-bus processing lives on the mix but not the stems, so the sum sounds different.
  • Cutdowns that run long: a 30 that lasts 31.2 seconds isn't a 30.
  • Fades instead of endings: a cutdown should end like music, not like someone turned it down.
  • Alts made months later from a different session: subtly different balance, tempo map or sounds.
  • Names nobody else can read: Mix 4 FINAL (2).wav.
  • Mixing up versions and splits: splits are ownership; send them on the metadata sheet, not as audio.
  • Delivering everything to everyone: a pitch email with 14 attachments gets skipped. Main mix, instrumental, "stems available on request".

The free checklist

To make this easy to apply, we put the whole delivery routine on two pages: what to print from the session, the pre-flight checks, the naming table and the metadata sheet columns. It's the free Sync-ready delivery checklist and file-naming standard (PDF and spreadsheet).

Download the Sync-ready delivery checklist and file-naming standard

Every version in one place, with Aerie

In Aerie, every version of a track lives under that one track: the main mix, alts, instrumental, cutdowns and stems, each with its own ISRC if you want one, assigned in bulk from your registrant code. Replacing a mix keeps the old one in its history, and the track's Sync Passport holds the splits, contacts and paperwork, so the right files and the right details go out together.

Frequently asked questions

What is the difference between stems and alt mixes?

Stems are separate layers of the mix (drums, bass, strings) that play back together as the full track. Alt mixes are complete, finished mixes with something removed or changed, such as an instrumental or an underscore.

Are alts, stems and splits the same thing?

No. Alts and stems are audio files. Splits are the ownership percentages of the song and the recording that you agree with your co-writers and publishers and register with your society.

What format should stems be delivered in?

Usually WAV or AIFF at 48 kHz, 24-bit, every stem starting at the same point and exactly as long as the main mix. Each library's delivery sheet has the final word.

Does a WAV file hold metadata?

Yes, in INFO, BWF (bext), iXML or ID3 chunks, but many programs don't read all of them. Always send a metadata sheet and use clear file names as well.

Should I make 30 and 60 second edits before pitching?

Usually not. Pitch with the main mix (and instrumental for songs), then make the cutdowns the library or client actually asks for, at their lengths.

Sources

More guides