Production guides · Meme Plugins · 26 October 2026
Match the loudness of the processed and bypassed signal before deciding whether a plugin improves the sound. Keep the same passage and listening volume, adjust the processed path’s output level, then switch between the two. Judge the intended change in tone or dynamics after the level jump is small enough to stop dominating the comparison.
This is a test of one processing decision. For a comparison between two finished songs, use the reference-track guide.
1. Decide what the plugin should improve
Choose a short, repeatable passage that exposes the problem. For an EQ, that might be a vocal line that becomes harsh on one vowel. For a compressor, include both a quiet phrase and a stronger entrance. Write a specific question: “Are the words clearer?” is easier to answer than “Is this more professional?”
Keep an unchanged session version. Leave speaker or headphone volume alone during the comparison, and listen at a comfortable level.
2. Find a gain control that belongs to the processed path
Use the plugin’s output or makeup-gain control when appropriate. Check what its bypass button actually bypasses. A trim placed after a plugin on the same track may remain active when that plugin is bypassed, turning down both versions instead of compensating only the processed one.
If you use two routes, give only the processed route its compensation gain and listen to one route at a time. Keep their routing and timing equivalent. Two copies playing together are a different experiment.
FabFilter’s Pro-Q output documentation describes output gain and its internal global bypass. It also explains that Auto Gain estimates compensation from EQ settings rather than measuring the programme’s actual loudness. Treat automatic gain as a starting point, then check the result yourself.
3. Match the passage, not just its tallest peak
Alternate processed and bypassed playback of the same phrase. Reduce the processed output if it seems louder; increase it if it seems quieter. A loudness or average-level meter can help you approach a match, but listen as well. EQ can change perceived brightness, while compression changes the relationship between attacks and sustained sound.
Identical sample peaks do not establish equal loudness. The meter guide explains that distinction. Use comparable measurement windows and do not compare one version’s whole-track reading with the other version’s chorus.
Example: an EQ boost with its volume advantage removed
Imagine an eight-bar guitar passage. A broad EQ boost makes the processed passage seem louder. In this hypothetical example, matching the passage requires about −1.5 dB of output trim. That number is an illustration, not a measurement or a rule that a particular EQ boost needs −1.5 dB.
- With compensation disabled, note whether you mainly hear an increase in level.
- Apply the matched output trim and compare again. Listen for pick attack, harshness and the guitar’s place beside the vocal.
- Keep the EQ only if its intended contribution survives the fairer comparison. Try a smaller boost if the new brightness distracts from the line.
Do not automatically subtract the EQ band’s gain from the output. Boosting one frequency region does not raise all of the signal by that amount.
4. Check the full chain and arrangement
Return to the full mix. A gain change before a later compressor or saturator can change that processor’s behavior. Decide whether you are judging the EQ alone or the complete chain, and place the comparison around the intended scope. Investigate overload with the clipping workflow.
Level-matching questions
Must the readings be exactly identical?
The aim is a useful listening decision. Get close enough that a level jump does not settle the vote, then listen for the change you meant to make. One meter cannot guarantee identical perception.
Should I keep the louder version afterward?
You can choose its final mix level separately. First decide whether the processing helps; then set the part’s level in the arrangement.
What if bypass clicks or changes the timing?
Use the processor’s documented comparison method and allow it to settle before judging. A switching artifact is not evidence that one EQ setting sounds better. Preserve latency compensation where the software requires it.