How to calculate line distribution fairness

What “calculating line distribution” means

A line distribution is a timing map of who sings when. Calculation usually means two related jobs: measuring each member’s singing duration, then summarizing those durations as shares and a fairness score. Fans, editors, and rehearsal coaches care about the summary because it answers a practical question—does one voice dominate the timed sections, or are the parts relatively balanced for the window you marked?

Duration can come from hold-to-time buttons while audio plays, or from typed ranges such as 0:12.0–0:18.5. Overlaps are normal in group vocals. A careful calculation therefore uses each member’s union of intervals rather than naively summing every raw segment and double-counting the same wall-clock second for one person. Shares on this site divide each member’s union duration by the sum of all member unions. Fairness then looks at those positive durations with Jain’s formula.

You do not need a statistics background to follow the worked scenarios. We will keep the algebra explicit, show intermediate sums, and round the final percentage the way a maker UI typically does. If you prefer to skip the arithmetic, open the free tool on the home page, mark demo members, and watch the live fairness percentage update as you edit.

The Jain fairness formula used here

Fairness here uses the Jain fairness index. Collect the positive singing durations x₁…xₙ for members who actually have timed lines. Members with zero duration are excluded from n. If fewer than two members have positive duration, this product shows “--” instead of a percentage, because a fairness comparison is undefined for a single singer or an empty set.

(Σx)² / (n · Σx²)

The result lies between 0 and 1. Equal positive durations approach 1 (displayed near 100%). When one duration dwarfs the others, the index falls. Display rounding may show 96% even when the exact fraction is 0.9615. That rounding is for readability; the ranking of members still follows the underlying durations.

Important scope notes: Jain fairness is not a moral judgment about a songwriting choice, and it is not a claim about stage presence, dance breaks, or visual screentime. It only summarizes the timed singing durations you marked. Changing the timed window—for example excluding an intro—changes both shares and fairness. Always treat the score as relative to the ranges you chose.

Scenario 1 — Even shares (demo)

Imagine a four-member demo project with fictional names Aria, Bora, Cleo, and Dina. After timing a short chorus window, each member’s union duration is exactly 10.0 seconds. No overlaps occur in this first scenario, so each union equals the single segment you marked for that person.

Σx = 10 + 10 + 10 + 10 = 40. Σx² = 100 + 100 + 100 + 100 = 400. n = 4. Jain = 40² / (4 · 400) = 1600 / 1600 = 1.0 → about 100% fairness. Shares are each 10 / 40 = 25%.

An even scenario is a calibration check. If your maker cannot reach roughly 100% when every positive duration matches, the implementation or the export you are reading is wrong. Equal demos also help new users trust that the UI is not hard-coded to a “balanced” look—fairness should fall as soon as you lengthen one member’s lines.

In practice, perfectly equal songs are uncommon, but rehearsal timers and practice stems sometimes are. Use an even demo when teaching a friend how the number works, or when verifying that CSV summary rows match the on-screen percentages before you build a spreadsheet chart.

Scenario 2 — Lead-heavy split (demo)

Keep the same four fictional members. This time the unions are Aria 22.0s, Bora 8.0s, Cleo 6.0s, and Dina 4.0s. The total of unions is still 40.0s so percentages stay easy to read, but the distribution is clearly lead-heavy: Aria alone holds 55% of the timed singing.

Σx = 22 + 8 + 6 + 4 = 40. Σx² = 484 + 64 + 36 + 16 = 600. n = 4. Jain = 1600 / (4 · 600) = 1600 / 2400 ≈ 0.667 → about 67% fairness.

Compared with the even case, the numerator stayed 1600 because the sum of durations did not change, but the sum of squares rose from 400 to 600. That increase is the mathematical signature of unevenness: squaring amplifies large values. Shares become 55%, 20%, 15%, and 10%. A lead-heavy result is not “wrong”; it may match how you timed a bridge featuring one voice. The fairness percentage simply makes that imbalance visible without reading every bar.

When editors argue about whether a cut “feels fair,” a lead-heavy score gives a shared number. You can still decide the artistic edit is correct. The calculation only answers whether the timed seconds are concentrated. If you later trim Aria’s harmony doubles from the timeline, both her share and the Jain index will move without you rewriting a spreadsheet by hand.

Scenario 3 — Overlaps between members (demo)

Overlaps are where hand-built spreadsheets often go wrong. Suppose wall-clock time runs from 0:00.0 to 0:20.0 in a demo. Aria sings 0:00–0:12, Bora sings 0:08–0:18, and Cleo sings 0:14–0:20. Dina has no lines in this window. Raw segment lengths are 12s, 10s, and 6s, which sum to 28s—but the wall-clock span is only 20s because Aria∩Bora and Bora∩Cleo overlap.

For fairness and shares on this site, each member contributes their own union duration: Aria 12s, Bora 10s, Cleo 6s. Dina is excluded because her duration is zero, so n = 3. Σx = 12 + 10 + 6 = 28. Σx² = 144 + 100 + 36 = 280. Jain = 28² / (3 · 280) = 784 / 840 ≈ 0.933 → about 93% fairness.

Notice that the sum of member unions (28) exceeds the wall-clock length (20). That is expected when people sing together. Shares use 12/28, 10/28, and 6/28—not 12/20—so the chart does not pretend overlapping voices fit into a single exclusive timeline. If you instead forced every second to belong to only one member, you would be calculating a different product: a mutually exclusive allocation, not the singing-time unions this maker reports.

Worked overlap demos also clarify Finish-lock and range editing. If you accidentally leave a duplicate segment for Aria covering the same 0:00–0:12 twice, a naive sum would inflate her duration, but a union pass should still report 12s. Checking Jain after cleaning duplicates is a fast sanity test before you export CSV for a spreadsheet template.

How the free maker applies these steps

On the home page tool, add members, mark timing pads or ranges, and read the live donut, ranked list, and fairness percentage. Exports include PNG for social posts plus CSV and JSON for archives. The CSV summary table mirrors the same durations and shares the UI uses, which is why the spreadsheet guide walks through importing those columns into Excel or Google Sheets without a login download pack.

Because projects can autosave in this browser on this device, you can refine overlaps, undo a mistaken hold, and re-check Jain without rebuilding the song. Local audio stays in the browser for the core flow. YouTube may be used as a reference player only. Nothing about the fairness math requires an account.

If you arrived from a Perchance-style generator search, the comparison page explains persistence and phone export. This fairness guide stays focused on calculation. For charting the exported numbers, continue to the spreadsheet template guide linked from the footer and below.

Read the line distribution spreadsheet template guide

Fairness FAQ

Do I need real group data to learn the formula?

No. Illustrative demos are enough to understand Jain fairness, and using fictional names avoids implying official statistics for any act.

Why did fairness drop when I added a harmony overlap?

Overlaps change each member’s union and the sum of squares. Fairness can rise or fall depending on whether the new intervals make durations more or less even—not merely because people sang together.

Is 70% “bad”?

The index is descriptive. A lead vocal spotlight can correctly land near the mid range. Use the number to communicate, not to grade artistry.