How I Prepare Mixes for Mastering in 2026: Exact LUFS, Stem Targets and Export Rules

Horia Stan is a music producer and sound engineer at The One Records, Bucharest.

I used to manage mix revisions the hard way. I accepted vague briefs, supplied single WAVs peaking at -0.5 dB, and waited for a mastering return that begged for stems. That changed in 2024 when streaming normalization and aggressive codec encoders made ambiguous mixes expensive to fix.

I run a strict mix-for-mastering discipline now. It saves time. It reduces revisions. It improves final translation across streaming platforms.

Why usual advice is broken in 2026

Most articles still tell you to leave "-6 dB headroom" and export a single file. That is lazy and inefficient in 2026. Streaming loudness normalization now centers around -14 LUFS integrated on major platforms. Encoders are harsher. Mastering engineers expect prepared stems with consistent loudness so they can reconstruct and correct balance fast.

Here's my claim: deliver a full mix set to -14 LUFS integrated with a true peak ceiling of -8 dBFS, plus a standard set of 6 stems measured to specific LUFS targets, and your mastering chain will take half the time and cost. I do not negotiate the numbers.

The exact numbers I use

• Full mix: -14 LUFS integrated, true peak -8 dBFS max. No final limiter on the master bus. No dithering.
• Stems: six stems exported at the same cadence as the full mix. Target integrated LUFS per stem: Vocals -18 LUFS, Drums -16 LUFS, Bass -20 LUFS, Instruments -20 LUFS, Backing Vocals -24 LUFS, FX -24 LUFS.
• File format: 24-bit WAV, 48 kHz. Name files: SongTitle_Stems_VOCALS.wav etc.
• Mastering safety: final mastering chain will use a ceiling of -1 dBTP for delivery masters. My mix does not attempt to hit that.

I measure LUFS with Youlean Loudness Meter 2 Pro on the mix bus and NUGEN VisLM when I need cross-checks. NUGEN MasterCheck Pro helps me preview how an encode will shift perceived loudness on Spotify and Apple.

Why these numbers, not the generic -6 dB advice

-14 LUFS integrated positions the mix within current streaming normalization defaults. If you hand over a mix at -8 LUFS integrated, the mastering engineer will either attenuate or push it into a limiter, which kills dynamics and increases artifacts after codec encoding.

A true peak ceiling of -8 dBFS gives mastering space to process without brickwall limiting during stem summing. Most mastering engineers prefer -6 to -8 dB. I pick -8 to force more headroom and avoid surprise inter-sample peaks when stems are summed.

Per-stem LUFS targets remove guesswork during mastering. If vocals are a defined -18 LUFS and drums -16 LUFS, the mastering engineer immediately sees the balance and can make targeted changes. This prevents the most common revision: "Make the vocals louder without touching the mix." That request becomes trivial when the vocal stem lands precisely where it should.

Exact workflow in Logic Pro (my setup)

Mix bus

1. Keep the master bus passive. Only a metering plugin. Insert Youlean Loudness Meter 2 Pro last. Do not use a limiter or brickwall processor on the master.
2. Set output channel fader so the mix peaks at -8 dBFS true peak. Watch Youlean for integrated LUFS - adjust fader automation to land at -14 LUFS integrated over the whole song.

Stems

3. Create six submix buses: VOCALS, DRUMS, BASS, INSTRUMENTS, BACKINGS, FX.
4. Bus only the relevant channels. Print effects that must remain (reverb/sends) to the stem. Do not print any master bus processing.
5. Use a separate instance of Youlean on each stem bus to measure integrated LUFS while the track plays. Adjust stem bus faders to hit the LUFS targets I listed.
6. Consolidate and export all stems and the full mix as 24-bit WAV, 48 kHz. Turn off dithering. Include a 20 ms fade in and 20 ms fade out to prevent clicks during stem summing.

Naming and metadata

7. Use strict filenames: Artist_Song_StemName_24bit_48k.wav. Example: HoriaStan_Liminal_VOCALS_24bit_48k.wav. Include a single-line README.txt with the LUFS numbers and the sample rate/bit depth.

What to avoid (and why I say it bluntly)

• No master bus limiting. Your limiter hides balance issues. It also produces encoder artifacts after streaming normalization.
• No dithering. Mastering should dither at final delivery. Dither in the wrong place is irreversible.
• No ambiguous stem exports. If you deliver 30 stems with ad-hoc levels, you will get charged for tedious balancing.

How this saves time and money in 2026

Mastering engines and services are faster but also more algorithmic. An engineer can run an AI loudness pass and a codec preview in 10 minutes. The manual work is balancing and creative EQ decisions. Good stem discipline collapses 3-4 rounds of revisions into one. I consistently cut mastering time by 40 percent with this method.

I saw the improvement across projects for clients like Andreea Bostanica and Edward Sanda. When stems arrive with predictable LUFS, I finish masters faster and with fewer artifacts after AAC/Opus encoding.

Edge cases and how I handle them

If the song has extreme low-end energy, I isolate sub-bass into a dedicated stem and target -22 LUFS for that stem. If the arrangement is vocal-forward electronic pop, I set vocals to -16 LUFS instead of -18 LUFS. Those are exceptions, not defaults. I choose them before export.

Step-by-step checklist

Final concrete takeaway

From now on, export one full mix at -14 LUFS integrated and true peak -8 dBFS, plus six stems with the LUFS targets I listed. Use Youlean Loudness Meter 2 Pro and NUGEN MasterCheck for verification. Deliver 24-bit/48 kHz WAVs, no dithering, clear filenames and a one-line README with your LUFS numbers. Do this once and mastering becomes a predictable, fast step instead of a money pit.