Prepare Stems for Mastering: 10 Simple Steps

Preparing stems for mastering means grouping your mix into 4 to 8 stereo bus exports, each starting at exactly 00:00:00, at 24-bit or 32-bit float WAV, with no dithering and enough peak headroom for the mastering engineer to work without clipping.

That’s the whole process. Everything below is how to do it without making the mistakes that send your session back to you with a polite but annoying email asking you to redo everything.

Step 1, Lock Your Mix First

Stem mastering doesn’t fix a bad mix. We’ve seen producers skip straight to stems because they think the mastering engineer will sort out the low-mids. They won’t. That’s your job.

Before you touch an export setting, play back your full stereo mix and sign off on it. Every arrangement decision, every EQ curve, every automation pass needs to be done. Stems are a snapshot of your mix at this exact moment.

If you’re still moving faders, you’re not ready to bounce stems.

Step 2, Decide How Many Stems You Actually Need

The industry accepts up to 8 stereo stems. But more stems means more cost and more complexity. Here’s how we approach it by genre.

For hip-hop and trap: 5 stems works well. Drums, bass, melodic instruments, lead vocal, and background vocals cover everything a mastering engineer needs to separate.

For electronic music: 6 stems. Kick and sub on their own bus, percussion and cymbals separate, synth bass, leads, pads, and FX atmospheres.

For rock and live band recording: 4 stems. Drums, bass, guitars and keys together, vocals. Unless you’ve got a complex orchestration, more splits just add noise.

Don’t pay for 8 stems if 5 covers your music. It’s that simple.

Step 3, Set Up Your Stem Buses Correctly

Create a bus channel for each stem group. Route all relevant tracks to that bus and nowhere else. The bus must not feed your master fader.

This is where most people get it wrong. If your stem buses also feed the master, you’ll export the stem AND hear it doubled in your stereo mix during playback. The buses need to be independent export paths only.

In Ableton: use return tracks only for time-based effects (more on that in Step 5). In Logic: use Aux channels routed directly to your output but soloed during export. In Pro Tools: use the Bounce to Disk from each bus individually.

Step 4, Handle Your Reverb and Delay Sends Carefully

This is the part nobody explains properly, and it’s frustrating to get wrong.

If you have a shared reverb send that three different instruments feed into, say, a room reverb on snare, piano, and guitar, you have two options.

Option A: Print the reverb wet signal into each stem proportionally. Solo the reverb return with only one group feeding it, and print that to the relevant stem. Repeat for each group.

Option B: Create a dedicated “FX Stems” bus that captures all shared returns together. Export it as its own stem. When the mastering engineer layers all stems back together, the FX stem adds those returns back in cleanly.

We use Option B. It’s cleaner, and it avoids the horrible phase issues you get when you try to print reverb tails into multiple stems and they overlap.

Step 5, Set Your Peak Levels

Your stems should peak no higher than -1dBFS. We prefer -3dBFS to -6dBFS peak level with no limiter on the stem bus.

Here’s why. We loaded a set of stems once that had a master bus limiter still running on each group bus. The kick was hitting -0.2dBFS with 4dB of limiting baked in. The mastering engineer had no transient left to work with. The snare sounded like someone hitting a cardboard box.

Pull the limiter. Pull the clipper. Remove anything on the bus that’s there purely for loudness.

Keep compression that is part of the sound. If your drum bus compressor is gluing your kit and you built the mix around it, leave it. If it’s a limiter you slapped on at mixdown to stop clipping, remove it and gain down the fader instead.

Step 6, Remove Master Bus Processing

Go to your master fader and bypass everything. All of it.

EQ, compression, saturation, limiting, if it’s on the master bus, it should not be on your stem exports. The mastering engineer has their own chain. Baking yours in gives them no room to do their job.

We keep a “Mastering Reference” channel strip saved as a preset, bypassed. Exporting is the moment you bypass it and leave it alone.

Step 7, Export at the Correct Bit Depth and Sample Rate

Export WAV at 24-bit minimum. Use 32-bit float if your DAW offers it for intermediate stages, it preserves calculation headroom better for the mastering engineer’s gain staging.

Keep the sample rate identical to your session. If you recorded at 48kHz, export at 48kHz. Do not upsample to 96kHz thinking it will sound better. It won’t. It just confuses the mastering engineer’s session setup.

Do not apply dithering. Dithering is for final delivery only, and the mastering engineer applies it at the end of their chain to the finished master.

Step 8, Check Sync: All Stems Start at 00:00:00

This is the most satisfying check in the whole process because it’s easy to verify. Load all your exported stems into a new session and stack them.

They should play back and sound identical to your original mix. If they don’t, something is routed wrong or an effect is missing from a stem.

We use SPAN by Voxengo on a sum bus in the verification session. A/B the stem sum against the original stereo mixdown. The frequency plot should be nearly identical within 1dB across the spectrum. If you see a 6dB dip at 80Hz, a bass bus didn’t export cleanly.

Step 9, Name and Organise Your Files

Use this naming format: SongName_StemName_BPM_Key_SampleRate_BitDepth.wav

Example: Silverline_Drums_128bpm_Amin_48k_24bit.wav

It sounds tedious. It’s not. The mastering engineer opens your folder and immediately knows what everything is without listening to 8 files to figure out which one has the kick in it.

Put all stems in one folder. Include a PDF or text file with your mastering brief: target loudness (LUFS), reference tracks, any specific notes about the low end or vocal brightness, and your intended platform (streaming, vinyl, sync).

Step 10, Run a Pre-Delivery QC Pass

Before you send anything, run this checklist.

Load all stems in a new session. Check: do they sum to your stereo mix? Check: do all files start at exactly 00:00:00? Check: do any stems clip above -1dBFS? Open iZotope RX’s level statistics on each file. Check: is dithering absent? 24-bit or 32-bit float files won’t have noise-shaped dither applied.

Listen to each stem in solo. You should hear a coherent group. If your “Instruments” stem has a stray vocal somewhere in it, fix it now.

Send a brief message with your delivery. One paragraph. Tempo, key, target platform, reference track, and one sentence on what you’re going for sonically.

Pro Tips

Use a 32-bit float bounce for all intermediate renders inside your session. Only drop to 24-bit for the final delivery files. This keeps calculation errors from stacking across multiple renders.

If you’re unsure whether processing belongs on a stem bus, ask yourself: “Would removing this change the character of the sound?” If yes, keep it. If you’re just using it to control level, remove it and gain down instead.

For stem mastering services, expect to pay 30% to 70% more than a standard stereo master. A stereo master might run $50 to $150 per song at a mid-tier studio. Stems for the same song often run $80 to $250. Decide if the extra control justifies the cost based on your release size. For a 10-track album, it often does. For a one-off single, a great stereo master is usually enough.

Keep a stem template session in your DAW. Pre-routed buses, correct export settings saved, blank mastering brief document included. The next time you need stems, the setup takes 10 minutes instead of 90.

Common Mistakes to Avoid

Leaving the limiter on the stem bus is the most common mistake we see. It’s ugly to fix after the fact. The mastering engineer receives crushed transients and has nothing to work with. Remove it before you bounce, always.

Exporting stems at a different sample rate than your session. If your session is 44.1kHz and you export at 48kHz, you’ve introduced a subtle pitch shift. Everything plays back slightly sharp. It’s subtle enough that some engineers won’t catch it immediately, but brilliant engineers will, and they’ll send the files back.

Forgetting to include reverb tails in the stem. Your drums stem needs to include the full reverb tail at the end of the song, not just the last hit. Export a few extra bars of silence after the final audio event to capture the full tail.

Sending stems without a brief. It’s not the mastering engineer’s job to guess what you want. “Warm low end, not too bright on the top, reference: [track name]” takes 30 seconds to write and saves a revision cycle.