When an artist or producer delivers the files for mastering, one of the most frequent mistakes has nothing to do with the mix itself: it has to do with how the files are exported and organized before sending them. A stem with clipping, misaligned with the rest or exported as MP3 can force the whole process to be repeated, or worse, produce a defective master that no one notices until it's already released.
This guide explains what stems are, when it makes sense to use them instead of a conventional stereo mix, and exactly how to prepare them so the mastering engineer can work without friction.
What are stems?
A stem is a partial mix of your project: an audio file that groups a subset of related tracks. Instead of exporting a hundred individual tracks (each drum mic separately, each synth layer, each vocal take), you export the big functional groups of your song.
The most common stems are:
- Drums and percussion. The whole kit together: kick, snare, hi-hats, overheads, additional percussion. There's no need to separate each drum element unless the engineer specifically asks.
- Bass. The electric bass, the bass synth, or both if they coexist in the song.
- Harmonic instruments. Guitars, keyboards, pianos, pads —sometimes grouped together, sometimes in two separate stems if they're very different in character.
- Lead vocal. The lead vocal, with its effects already integrated if they're part of the final sound.
- Backing vocals and doubles. Backing vocals, harmonies, vocal doubles.
- Effects and specials. Ambience elements, foley, loose samples, creative FX that don't fit the previous groups.
The key difference between stems and multitracks is that grouping. The multitrack is all the individual tracks unprocessed or with minimal per-track processing: it's the raw material of the recording. Stems are the result of having partially mixed those groups in the DAW —with EQ, compression and effects applied— but without having combined them yet into the final stereo. They're something like the midpoint between the multitrack and the finished mix.
Stems vs stereo mix
When we talk about mixing vs mastering, the difference is clear: mixing works on individual tracks and mastering works on the finished mix. Stems add a third, intermediate option, known as stem mastering.
The practical question is when that extra step of complexity is worth it.
The conventional stereo mix is simpler to prepare, faster to transfer and enough for most projects where the mix is already well resolved. The mastering engineer works on a single file and applies their processing chain to the whole. It's the standard flow for the vast majority of releases.
Stems add control when there's something in the mix balance that isn't quite resolved and would be hard to correct by acting only on the final stereo. If the vocals are slightly low relative to the instruments, or if the bass has too much presence at a specific frequency that affects the rest, the engineer can adjust that stem group before going through the mastering chain. It's not a substitute for mixing —it's still better to arrive with a well-balanced mix— but it does give a margin of correction the stereo doesn't allow.
In short: if your mix is good, the stereo mix is enough. If there's something that doesn't quite work in the balance between groups and you don't want to reopen the mix session, stems are the alternative.
Stem export checklist
This is the technical part that generates the most mistakes. Follow these points before sending any file:
| What to check | How to do it correctly |
|---|---|
| Headroom | Each stem should have a peak of approximately -6 dBFS. Check the peak level with your DAW's meter before exporting. If any stem exceeds -3 dBFS peak, lower the corresponding bus fader before exporting. |
| No limiter on the master bus | Disable or remove any limiter, maximizer or compressor you have on the master channel before exporting the stems. The master bus must be clean: no processing that affects level or dynamics. |
| Same start point | Export all stems from exactly the same zero point of the project (bar 1, beat 1, or the absolute start of the session). If each stem starts at a different point, the files won't line up when imported. |
| WAV format, 24-bit minimum | Export as WAV (or AIFF). Bit depth: 24-bit minimum; 32-bit float if your DAW works internally at that resolution. Never MP3 or any lossy format. |
| Original sample rate | Keep the same sample rate you recorded and mixed at (44.1 kHz, 48 kHz, 96 kHz…). Don't resample on export: resampling introduces subtle changes in the audio that can accumulate. |
| Clear naming | Name each file with the song name and the group: Song_Drums.wav, Song_Bass.wav, Song_Vocals.wav. Avoid generic names like stem1.wav or export_final.wav. |
| Reference mix | Also include an export of the full stereo mix (with the master bus clean, no limiter). It's the reference the engineer uses to know how you wanted the balance between groups to sound. |
Check each point before compressing and sending the files. A poorly exported stem can invalidate all the work that follows.
Typical mistakes when sending stems
Clipping on one or more stems. It's the most frequent and most damaging mistake. If the drum stem reaches 0 dBFS or exceeds it, the signal is digitally clipped: that distortion is already recorded in the file and has no possible correction in mastering. Always check the peak level of each stem before exporting.
Misaligned stems. This happens when each stem is exported from a different start point, or when the "export selection" function is used instead of exporting from the absolute start of the project. When imported into the engineer's DAW, the groups don't match in time and the song sounds out of sync. The solution is always to export from the same zero point for all files.
Embedded master-bus processing. If you had a limiter active on the master bus when exporting, that processing is already inside the audio: the WAV file sounds limited even though the file technically isn't clipping. The engineer receives material with the dynamics already crushed, with no room to work. Remove all master-bus processing before exporting.
MP3 instead of WAV. MP3 uses lossy compression: it discards audio information to reduce file size. In casual listening it may not be noticeable, but in mastering —where EQ, compression and limiting corrections are applied— the compression artifacts are amplified and can ruin the result. The space saving doesn't justify the risk: always use WAV.
Incorrect or resampled sample rate. Exporting at a sample rate different from the project's —for example, converting from 48 kHz to 44.1 kHz before sending— introduces changes in the audio and can cause alignment problems if the files are imported without conversion. Always send at the project's native sample rate.
How to deliver the files once exported
You have the stems exported and checked. Now you need to package and send them so the engineer can import them without friction.
Compress them into a single ZIP with a descriptive name. Put all the stems plus the reference stereo mix into a single folder and compress it. Use a name that identifies the project unambiguously: ArtistName_SongName_Stems.zip. Avoid generic names like files.zip or stems_final_v2.zip that say nothing about the content.
Use a service made for large files. WAV stems are heavy: a four-minute song with six stems at 24-bit/48 kHz can easily exceed 500 MB. Don't send them by email or compress them to MP3 so they "fit". The usual options are WeTransfer (up to 2 GB free, no account), Dropbox or Google Drive. Any of the three works well; what matters is that the link is accessible and the files are in their original WAV format without recompressing.
Include a note with the project context. Don't assume the engineer knows how your song sounds from the files alone. Along with the ZIP, or in the same delivery message, state: the track's BPM, the key if you know it, the sample rate and bit depth you exported at, and one or two sonic references —commercial songs you like that serve as a sound target. You don't need an extensive dossier: four lines are enough and save a round of questions.
Don't touch the files after exporting. Once you have the stems correctly exported, don't rename them, don't reorder them and don't open them in another editor "to listen". Any accidental save can alter the metadata or, in the worst case, the audio. If you spot an error after exporting, go back to the DAW, fix it there and export again from scratch.
What NOT to include in the stems
There's a group of processes that belong exclusively to mastering and shouldn't already be applied on the stems you send:
Limiter or maximizer on the master bus. Already mentioned, but worth repeating because it's mistake number one: any limiting or maximizing plugin on the master channel must be disabled on export. That includes iZotope Ozone, FabFilter Pro-L, Waves L2, or any other. If you have it as a reference for how you want the final result to sound, describe it in the notes you send the engineer, but don't leave it active on the export.
Normalization. Don't normalize the files before sending them. Normalization raises or lowers the file's level so the peak reaches a specific value, but in the process it can destroy the headroom the engineer needs. Export at the real bus level without touching normalization.
Premature dithering. Dithering —the technique of adding controlled noise when lowering the bit resolution— is only applied in the final mastering step, when the file is exported to 16-bit for distribution. If you apply dithering when exporting your stems to 24-bit, you're adding unnecessary noise that accumulates in the following steps. Most DAWs don't add dithering by default when exporting to 24-bit, but check the export settings just in case.
Frequently asked questions
Should I send stems or the stereo mix?
Stems give the mastering engineer more control: if the balance between groups isn't perfect (for example, the vocals are too low relative to the instruments), they can correct it before mastering. The stereo mix is simpler to prepare and is usually perfectly enough when the mix is already well resolved and you don't need that extra room for correction.
What sample rate and bit depth should I use for stems?
Use the same sample rate you worked with throughout the session —don't resample on export—, and 24-bit as a minimum bit depth. If your DAW works in 32-bit float internally, you can export in 32-bit float without issues. The format must be WAV or AIFF. Never send MP3: lossy compression introduces artifacts that get amplified during mastering.
Ready to send your project?
If you already have the stems prepared or want to confirm everything is in order before exporting, write to us. We'll review the files and tell you if there's anything to adjust before starting.
Get a quote