How to Prepare Stems for Mastering

Mastering can only be as good as the files it receives. A mastering engineer or AI processing engine cannot fix a clipped stem, recover headroom that was squashed by a limiter, or undo a sample rate conversion artifact. Getting your export settings right is a simple process that takes five minutes and prevents hours of frustration. This guide is part of our mastering and delivery hub, and it gives you a concrete, repeatable checklist for every export.

Whether you are sending stems to a human mastering engineer, uploading to an online mastering service, or processing your own master, the preparation steps are identical. Proper stem prep is a sign of professionalism and ensures the mastering stage can focus on creative refinement rather than damage control.

Stems vs a Stereo Mix: What to Submit

A stereo mix is a single WAV file of your entire song, left and right channels, printed as one file. This is the traditional format for mastering. The mastering engineer works with the balance you established in the mix and applies global processing (EQ, compression, limiting) to the whole signal.

Stems are grouped submixes exported as separate files. A typical stem set includes drums, bass, vocals, melodic instruments (synths, guitars, keys), and effects or production elements (risers, impacts, ambient textures). Each stem is a stereo or mono file that represents one category of your arrangement.

Stem mastering gives the engineer more surgical control. If the vocal needs 1 dB more presence, they can EQ the vocal stem without affecting the guitars in the same frequency range. If the bass is slightly too loud relative to the drums, they can adjust the balance without touching the rest of the mix. AI mastering services especially benefit from stems because the processing engine can analyze and optimize each frequency group independently.

The standard approach is to provide both: a stereo reference mix that shows the balance you intended, plus individual stems for mastering flexibility. Always include a stereo mix even if you are sending stems so the engineer knows what your vision sounds like.

Recommended Export Settings

These settings apply to every stem and your stereo reference mix. The goal is to deliver the highest fidelity version of your audio with no unnecessary processing baked in.

A quick note on AIFF vs WAV: both are uncompressed and lossless. WAV is the universal standard. Use WAV unless your mastering engineer specifically requests AIFF. Do not export as FLAC or ALAC for mastering; while these are lossless, some DAWs and processing tools handle them less gracefully.

Headroom: How Much Space to Leave

Headroom is the gap between your loudest peak and 0 dBFS (the digital ceiling). For mastering, you want your stereo mix to peak between -3 and -6 dBFS. Individual stems should sit slightly lower still, because when they are summed together in the mastering session, their combined level will be higher than any single stem.

Creating headroom is straightforward. If your mix is peaking at -0.5 dBFS, simply pull the master fader down by 3 to 6 dB before bouncing. This does not change the relative balance between elements; it just lowers the entire signal. Every internal balance, every send level, every automation move remains exactly the same. You are just giving the mastering chain room to breathe.

Do not confuse headroom with dynamics. Headroom is about peak level relative to the ceiling. Dynamics is about the difference between loud and quiet moments. You can have plenty of headroom and very little dynamic range (a heavily compressed mix turned down), or very little headroom with wide dynamics (an uncompressed mix peaking near 0 dBFS). For mastering, you want both: enough dynamics for the limiter to shape, and enough headroom so the mix translates cleanly through every processing stage.

What to Disable Before Exporting

Before you bounce stems or a stereo mix for mastering, audit your master bus and remove or bypass any processing that the mastering engineer will handle. This includes:

• Master bus limiter: The mastering engineer will apply their own limiter with precise settings. A pre-baked limiter removes headroom and dynamics that cannot be recovered.
• Master bus compressor: Unless it is a fundamental part of the mix's sound (and you mixed into it from the start), remove it. Heavy bus compression before mastering leaves little room for the mastering compressor to work.
• Master bus EQ: Remove surgical or corrective EQ from the master bus. If you applied a broad tonal EQ that you mixed into and the mix sounds wrong without it, leave it on but communicate this to the mastering engineer.
• Stereo wideners: Master bus stereo processing should be handled in mastering where it can be dialed precisely against the final loudness level.
• Loudness meters: Not harmful, but remove metering plugins from the signal path to avoid any latency or processing overhead during bounce. Route them on a separate bus instead.

Leave all processing on individual tracks and group buses intact. Those are part of your mix. Only strip the master bus and let the mastering stage handle global processing from scratch.

File Naming Conventions

Clean file naming saves time and prevents confusion. Use a consistent naming scheme for every project:

Use underscores or hyphens instead of spaces. Avoid special characters, accents, and excessively long names. Include the tempo and key in a separate text file or in the folder name if relevant. A mastering engineer receiving a folder named "JohnDoe_Sunrise_120BPM_Cmaj" with clearly labeled stems can start working immediately without a single email exchange.

How to Group Your Stems

There is no universal standard for stem grouping, but the following 5-stem structure works for most genres and gives the mastering stage enough control without overwhelming it with dozens of files:

1. Drums and Percussion: Kick, snare, hi-hats, toms, overheads, percussion loops, and any sample-based drum elements. Print with their individual processing and group bus processing intact.
2. Bass: Bass guitar, synth bass, 808s, sub-bass layers. Keep these on their own stem because low-frequency control is critical during mastering.
3. Vocals: Lead vocals, harmonies, ad-libs, vocal chops. Include time-based effects (reverb, delay) if they are baked into the creative sound. Optionally provide a separate dry vocal stem.
4. Melodic Instruments: Guitars, synths, keys, strings, brass, pads. Everything that carries melody or harmony outside of vocals and bass.
5. FX and Production: Risers, impacts, transitions, ambient textures, noise layers, reverse effects. These are the production elements that do not fit cleanly into the other categories.

Verify your stems by importing all of them into a new session and playing them back together. They should sum to match your stereo reference mix exactly. If the summed stems sound different (louder, quieter, or tonally off), something was exported incorrectly. This verification step catches mistakes before they reach the mastering signal chain and cause problems downstream.

Common Export Mistakes to Avoid

• Clipping on export: If any stem or the stereo mix clips (peaks above 0 dBFS), the distortion is permanent. Always check meters before bouncing and pull levels down if necessary.
• Wrong sample rate: Exporting a 44.1 kHz session at 48 kHz (or vice versa) forces a sample rate conversion that can introduce subtle artifacts. Match the export rate to your session rate.
• Missing tracks: Forgetting to solo a group before bouncing or accidentally muting an element. Always verify by summing your exported stems against the reference mix.
• Stems that start at different times: All stems must start at the same point in time (typically bar 1, beat 1 of the session). If one stem starts two bars later, the mastering engineer cannot align them. Export all stems from the same start and end point, including silence.
• Dithering on stems: Dither is noise shaped to mask quantization distortion when reducing bit depth. Apply it only once, at the very final conversion to 16-bit for CD distribution. Dithering on 24-bit stems adds unnecessary noise floor.
• Exporting as MP3: Never send MP3 files for mastering. MP3 is a lossy format that permanently removes frequency content. Even 320 kbps MP3 is not suitable for mastering input. Always use WAV.