The principle
When you compare two versions (before and after a plugin, two masters, your own track and a reference), the louder version almost always seems better: fuller, brighter, more powerful. It's an effect of equal-loudness contours (at higher volume we perceive more lows and highs) and of our instinctive preference for volume. Even 0.5–1 dB of difference is enough to skew judgment. To really decide, you have to compare at equal loudness.
How to match loudness
- Measure: use LUFS (short-term or integrated on the same passage) rather than peaks or judging by eye.
- Compensate: every plugin that changes the level must have its output adjusted so that bypassed and active have the same loudness. Many plugins have an “auto gain” / “gain match” / “output compensation” option (Ozone has a Gain Match function for comparison; some limiters offer gain-compensated listening).
- Comparison tools: ADPTR Metric AB, Mastering The Mix REFERENCE, MeldaProduction MCompare, iZotope Tonal Balance: they let you switch between sources with automatic loudness matching.
- In FL Studio you can put a Fruity Balance at the end of the chain to compensate the gain, measuring with a LUFS meter.
Review method
- Listen first, then measure: the ears decide, the meters check.
- Short switches (5–10 seconds) on the same passage.
- Blind whenever possible: ask someone to switch, or use the “blind” function of comparison plugins. An ABX test (identifying which of the two is X) tells you whether a difference is real.
- Multiple volumes: low, medium, (briefly) high.
- Multiple systems: monitors, headphones, phone, car.
- Breaks: the ear adapts in a few minutes; listen again the next day.
- Take notes: time, problem, hypothesis, action (“1:12 snare harsh at 3.5 kHz → dynamic EQ −1.5 dB on the bus”).
Reviews with the client or the artist
- Send files with comparable loudness (or specify that they are final-level masters).
- Ask for comments with timestamps and compare with agreed references.
- Number the versions (v1, v2…) and keep the settings (recall) for each version.
- Limit the revision rounds: 2–3 well-handled rounds beat 10 chaotic ones.
What to evaluate in a mastering A/B
| Aspect | Question |
|---|---|
| Tone | is it better balanced against the references? |
| Low end | clearer or just louder? |
| Vocal | more present without becoming harsh? |
| Transients | do kick and snare still have punch? |
| Stereo | wider image but a solid center? |
| Depth | has the space been preserved? |
| Distortion | are there limiter or clipper artifacts? |
Example
Two masters of a pop track in A minor: A with a 1.5 dB boost at 10 kHz and limiter at −8 LUFS, B with no boost at −9 LUFS. Heard like this, A seems better. Matched at −9 LUFS, A turns out to be only harsher on the sibilants; B has a more natural vocal. Blind test: 7 out of 10 prefer B. B is chosen and only +0.5 dB of air is added.
Common mistakes
- Evaluating a plugin by switching it on without compensating the gain.
- Switches that are too long (you forget the timbre).
- Trusting the first listen after hours of work.
- Reviews without precise notes.
Exercise
For a week, every time you insert a plugin, adjust the output for an equal-loudness bypass before judging it. Note how many times the plugin that seemed to “improve” things was really just raising the volume.
Links
“Reference Tracks”, “Psychoacoustics in the Mix”, “Loudness War and Perceived Dynamics”, “Loudness, LUFS, and Normalization”, “Goals of Mastering”.