Stem Energy Share
Balancing five stems by ear is guesswork with the reference track in the other room. What you actually need is a way to say out loud how much of the mix energy each stem carries, because fader positions and meters describe different things. A kick sitting at minus 4 dBFS and a bass at minus 12 dBFS sound far closer together than those numbers suggest, and a vocal at minus 15 dBFS is often already buried before anybody complains. This tool converts every stem level into linear energy, sums it, and reports each stem as a percentage of the total. That is the question that predicts whether someone listening on a phone speaker still hears the bass line. It also flags three classic failures: a low end where the kick does all the work, a vocal buried under the lead, and a midrange that has swollen into a wall. Then it compares your numbers against a reference balance for the genre you are working in.
How to use it
Mark the stems that actually exist. Absent stems are excluded from the energy total instead of being counted at minus infinity, which matters more than it sounds: a five stem pack and a three stem pack have completely different definitions of a healthy bass share. If a part is printed but silent, mark it absent. A silent stem marked as present adds nothing but confusion to the report.
Enter the levels in dBFS. Use the RMS or LUFS style reading for each stem rather than a single peak number, because a peak is dominated by the loudest transient and says nothing about how much energy the stem carries. Roughly, a stem that sits at minus 8 dBFS RMS is doing far more work than one at minus 8 dBFS peak. Write down the number you would defend in a session.
Read the share, not the fader. The percentage column tells you how much of the total energy each stem owns, which is the number that predicts what survives translation. A stem at minus 14 dBFS holding 6 percent of the energy is buried in a way no fader position can hide, and a stem at minus 4 dBFS holding 60 percent of it is carrying the entire record.
Work the findings in the order they are listed. Each fix names the stem to move first and the number of decibels to move it, starting with a small step so you can hear the change before committing to the full correction. Re-measure after every move, since the percentages recompute against a new total and one adjustment often fixes two findings at once.
Three common mistakes
- Mixing peak readings with RMS readings. A kick with a strong transient can peak at minus 1 dBFS while its average level sits near minus 12 dBFS, and feeding the peak number into this tool makes the kick look like it owns the whole mix. Measure every stem the same way, on the same meter, over the same passage of the track, or the shares are not comparable with each other.
- Treating the percentage as a prescription. The numbers describe an approximation of where the energy sits, not how the record should sound. A techno record with a kick at 60 percent of the energy can be perfectly balanced if the other 40 percent is still audible, and a house record with a heavy vocal will sit outside the table on purpose. Use the deviations to find what is hiding, not to chase numbers.
- Forgetting that stems overlap in the low band. Two stems carrying the 80 Hz region are not two separate buckets of energy, they are one region heard twice, so the real low end is louder than the sum suggests. This tool treats every stem as if it were broadband and uncorrelated, which is the only way to get a single comparable number, and it is why the low band share usually reads higher on paper than it sounds.
- Reading the low band without separating the sub. A bass stem at minus 12 dBFS that is all sub energy will still register as a healthy bass share here, and yet it will disappear on every speaker with no low end. Split the sub out of the stem, or check the low end with a small speaker and a headphone, before deciding that the balance is fine.
Energy, not faders: how these numbers are derived
Decibels are logarithmic, so adding them tells you nothing. The way to compare two stems is to convert each level into linear energy with 10 raised to the level divided by 10, then add those numbers. A stem at minus 4 dBFS has 0.3981 units of energy. A stem at minus 12 dBFS has 0.0631 units. The second stem is 8 dB quieter, which sounds like a small difference and is actually a factor of six. Summing the five stems gives the total energy, and each stem share is that stem divided by the total.
- The share is a proportion of energy, not a proportion of loudness. A stem can sit lower on the meter and still hold a large share if the other stems are lower still.
- The low band here means the kick and the bass stems, which normally cover the 60 to 120 Hz region and the sub content underneath it. The mid band means the lead and the vocal.
- Deviation is measured in percentage points against the genre reference, with a tolerance of eight points before a stem is called out.
The model is deliberately simple: every stem is treated as broadband and uncorrelated, so overlapping content in the low band is not modelled and a mono-summed mix will read differently from the real thing. That approximation is fine for spotting a kick that owns the low end or a vocal that has been left behind, and it is not a substitute for listening on a small speaker with the track playing.
The three classic failures and the moves that fix them
Kick dominance means the bass holds under 35 percent of the low band energy. The kick is then doing the entire job of the low end, the bass reads as a texture under it, and on a phone or a small system the low end collapses into a click with no body. The correction is arithmetic: the bass needs a specific number of decibels to reach a 35 percent split, and the same number can be taken off the kick instead. Start with half of that move, listen, and re-measure, because one adjustment often also opens up the mid band.
- Vocal masking is flagged when the lead sits 4 dB or more above the vocal. Half of that gap is usually enough to let the vocal breathe, and a dynamic EQ carved around 1.5 to 3 kHz can make the same change without touching the fader at all.
- Mid band crowding appears when the lead and the vocal together hold more than half the total energy. The fix is a small cut on whichever mid element is widest, not a cut on the vocal, which is usually the element the song needs.
- Very little air means the FX and texture stem is under 2 percent of the energy, which usually means the transients and the top of the arrangement are not doing anything.
Order matters. Fix the low end first, because every level decision below it is relative, then fix the vocal against the lead, then look at the top. Making the top brighter before the low end is settled will only make the imbalance more obvious.
Using the genre table without cargo culting it
The reference balances come from what tends to be heard on released records, and they are starting points rather than targets. Techno sits with the kick at 38 percent and a small vocal at 8 percent because the vocal is usually a guest feature. House moves weight to the vocal at 16 percent because the vocal carries the record. Drum and bass keeps the kick and the bass close together because half time pacing needs a balanced low end, and trap leans on the bass at 30 percent because the sub is the hook.
- Deviation is only interesting when it matches something you can hear. A lead 20 points over its reference that sounds fine is fine.
- Compare against the reference track in the same genre, matched in level, before you move anything. The table tells you where you are, not where you should be.
- Reference balances assume a five stem layout. A mix with a separate sub stem, or with a percussion stem, has a different denominator, and the percentages shift for that reason alone.
Use the table to ask better questions in the session. If the bass is 18 points under its house reference, ask whether the track is missing a bass line or whether the reference track simply has a louder one. If the vocal is 10 points over, ask whether the mix needs a thinner instrumental or whether the vocal part is simply too loud in the session.
FAQ
Why is the bass share so much lower than its dB level suggests?
What is the right kick to bass balance?
Should the vocal hold more energy than the lead?
How accurate is this, really?
My mix does not match the table and it sounds fine. Now what?
Limits of this tool
- The energy model treats every stem as broadband and uncorrelated, so overlapping content in the low band is not modelled and the real low end is louder than the sum suggests. Treat the shares as a comparison tool rather than a measurement.
- Levels are expected as RMS or LUFS style readings measured the same way on every stem, over the same passage of the track. Peak readings will skew the result badly, because a single transient can dominate a peak meter.
- Genre reference balances are starting points from common practice with an eight point tolerance, and they assume a five stem layout. A separate sub stem or a percussion stem changes the denominator and therefore every percentage.
- Nothing is uploaded, stored or logged. The report is assembled in the browser from the numbers you typed, and the copyable version can be pasted into any session note, email or tracker without leaving your machine.