Definition
In stem mastering the mastering engineer receives not a single stereo file but a few groups (stems) from the mix: for example drums, bass, music (harmony and melodies), vocals, FX. Summed without changes, the stems must rebuild the mix exactly. This allows corrections that would be impossible or costly on the stereo file (for example raising the vocal by 1 dB or reducing the kick without touching the bass), while keeping mastering's role: refining, not remixing.
When it makes sense
- The mix has one or two balance problems the producer can't solve in their room (often the low end).
- Electronic productions where the producer mixes alone and wants outside control over kick/bass.
- Projects with many versions (instrumental, a cappella, TV mix) that are derived directly from the stems.
- When the mix is not accessible (session from another studio) but the stems are.
When it doesn't
- If the mix is good: stereo mastering is simpler and preserves the mixing engineer's choices.
- If the problem is big: better to go back to the mix.
- If the mastering engineer is tempted to “remix”: costs and time go up, the identity of the mix is lost.
How to prepare the stems
- Number: 4–8 stems. Typical: DRUMS (or KICK separate + DRUMS), BASS, MUSIC, LEAD/SYNTH, VOX (lead), BGV (backing vocals), FX.
- Same start point and same length for all (from the start of the project, with complete tails).
- Effects included: reverbs and delays must be printed in the stem of the element they belong to (or supplied as a separate stem, “REVERBS”), otherwise the mix can't be rebuilt.
- Master bus: turn off compressors and limiters on the master bus before exporting, or provide the “pre” stems and the “post” stereo mix with a precise note of the master chain. If the sound depends on master compression (glue), tell the engineer.
- Sidechain: if the kick pumps the bass, the pumping stays printed in the bass stem (good); but if you remove the kick, the bass keeps pumping: point this out.
- Verification: import the stems into a new project, sum them at 0 dB, and compare them with the stereo mix by inverting polarity: you should be left with silence (or nearly, if the master bus had nonlinear processing).
- Format: 24-bit WAV or 32 float, project sample rate, no dither, no clipping.
- Reference: also send the stereo mix as you intend it.
In FL Studio
Route each group to a bus (DRUMS, BASS, MUSIC, VOX, FX) and make sure the reverb/delay sends are on consistent buses (e.g. a reverb return for the vocal that goes into the VOX bus, not the master). Then File → Export → WAV with Split mixer tracks, and use only the bus files. Alternatively, render each bus by soloing one group at a time.
The engineer's work
The engineer processes each stem with limited moves (EQ, light compression, saturation), then sums everything on a mastering bus with the usual chain. Typical corrections: ±0.5–2 dB of level between stems, EQ on kick and bass lows separately, vocal de-essing, control of cymbal brightness.
Example in A minor
Future bass track at 150 BPM: the producer exports KICK, DRUMS (without kick), SUB, SYNTH (supersaw and pluck), VOX, BGV, FX, REVERB/DELAY. The master reveals that the SUB covers the kick at 50 Hz on some notes and that the vocal is 1 dB low in the chorus: the engineer applies a dynamic EQ on the SUB at 50 Hz driven by the KICK and raises VOX by 1 dB in the chorus, then masters to −8 LUFS. The result stays faithful to the mix but solves the two problems.
Common mistakes
- Stems that don't sum to the mix (missing effects, different master bus).
- Too many stems (20 tracks): that's a mix, not mastering.
- Stems with individual limiters that change the dynamics.
Exercise
Export the stems of one of your tracks and run the inverted-polarity sum test. Then master from the stems and from the stereo file: in which places did the stems give you a real advantage?
Links
“Organizing the Session in FL Studio”, “Goals of Mastering”, “Formats and Delivery”, “Phase and Polarity”.