When I was trying to release my last single, I got stuck. Hours and hours of subtle eq moves and plugin choices to try and get it sounding just right. Toggling back and forth between various versions that were all starting to sound both exactly correct, and still not quite right. Sound familiar?

That was a mix — but the same thing happens the moment you run a podcast through noise reduction. Auphonic, Adobe Enhance, Descript: it sounds cleaner, so you ship it. Did it actually get better, or did you just trade hiss for a faint underwater warble and stop noticing?

The problem isn’t EQ or noise reduction specifically — it’s this: the moment you can see which version is “the processed one,” you can’t judge it anymore. Two biases stack up against you.

The two biases working against your ears

Loudness bias. Louder almost always sounds “better” to humans. Even a decibel or two of level difference is enough to tip your judgment — and most processing changes the level. So if your “after” is a hair louder, you’ll prefer it for reasons that have nothing to do with whether it’s actually cleaner.

Confirmation bias. You want the cleanup to have worked. You spent time on it. You know which one is the “fixed” version, so you hear it as fixed. That’s not a character flaw — it’s how perception works when you have an expectation.

Put those together and “I listened to both and the processed one sounds better” tells you almost nothing.

The fix: blind, level-matched A/B

The way out is the same method publications use when they pit noise-reduction tools against each other in a shootout: a blind, level-matched comparison.

  1. Match the levels. Comparison only means something if both clips are at the same perceived loudness. Otherwise you’re testing volume, not quality.
  2. Hide which is which. Label them A and B at random so you genuinely don’t know which is your raw take and which is your cleanup.
  3. Pick the one that sounds better. Commit before you find out.
  4. Then reveal. Sometimes the cleanup wins. Sometimes the raw take wins. Sometimes you can’t tell them apart at all — which is its own useful answer: if you can’t hear the difference, the processing isn’t earning its artifacts.

How to actually do it

You don’t need to set this up in your DAW. We built Blind Listen for exactly this — it runs in your browser, your files never upload anywhere, and it level-matches and randomizes for you:

  • Drop in your before (raw) and after (cleaned) of the same clip.
  • Listen blind. Pick the better one.
  • See the reveal — and find out whether your cleanup helped or just changed things.

It’s free and there’s nothing to install.

Beyond cleanup: blind-test the decisions you make on autopilot

Once you have a blind A/B habit, point it at the choices you currently make on faith:

  • Tool shootouts — Enhance vs Descript vs Hush vs your plugin chain, on your voice and your room, not a reviewer’s.
  • Settings — aggressive vs gentle noise reduction; how much de-reverb is too much.
  • Mics and chains — is the upgrade audible, or are you paying for a difference you can’t hear?

The goal isn’t to second-guess everything. It’s to stop shipping changes that feel like improvements but aren’t — and to spend your editing time on the ones that actually move the needle.

Trust the method, not the memory of which button you pushed.

Try Blind Listen — free, in-browser, nothing uploaded.