People forgive bad video. Bad audio? They just leave.

Picture two versions of the same YouTube video. Same person, same script, same slightly shaky webcam. In one version the voice is clean and close. In the other there is a soft hiss, a bit of room echo, and the level is low enough that you nudge your volume up. Which presenter knows their subject better?

Logically, neither. The words are identical. But that is not how listeners behave, and there is real research on this.

The experiment that should worry every creator

In 2018, researchers Eryn Newman and Norbert Schwarz ran a simple experiment. They took real conference talks and NPR science interviews and made two versions of each: one with the original clear audio, one degraded. Content identical, word for word. Then they asked people to evaluate the research and the researcher.

The degraded-audio versions lost on almost every measure. Participants rated the same talk as worse, the same speaker as less intelligent and less likeable, and the same research as less important. The study was published in Science Communicationunder a title that gives away the finding: “Good Sound, Good Research.”

The uncomfortable part is that listeners did not report noticing the audio. The judgment happened underneath. Psychologists call the mechanism processing fluency: when something is harder to process, we quietly hold it against the source, whatever the actual cause.

And this is not only a creator problem. The study was about perceived competence, which is exactly what is on the line in a remote meeting, a sales call, or a recorded presentation. If you pitch through a laptop mic from an echoey room, some fraction of your credibility is paying for it, and nobody in the meeting will tell you.

Audio and video are not judged equally

There is an older finding that deserves more attention than it gets. Research on perceived audiovisual quality going back to the early 2000s found an asymmetry: audio quality pulls down the perceived quality of the video, but video quality does not do the same to perceived audio. Your viewers experience your content as one thing, and the audio channel carries more of that impression than seems fair.

You have felt this yourself. A film scanned from grainy 16mm stock reads as a stylistic choice. A film where the dialogue crackles reads as broken. Podcasters listen through car speakers and cheap earbuds to content recorded on $80 microphones, and nobody minds, because what matters is not fidelity. It is the absence of problems.

What “bad audio” actually is, in numbers

Almost all of the audio that gets punished by listeners has one or more of four specific problems:

• Noise. Fan hum, hiss, street noise. Once the gap between your voice and the background shrinks below roughly 20 dB, listeners start working to separate the two, and you get billed for that work.
• Room echo.Untreated rooms smear every word with reflections. This is the “recorded in a bathroom” sound, and it is the hardest problem to fix after the fact.
• Low level. Quiet recordings force the listener to turn up, which turns up the noise too. Spoken content on major platforms typically plays back around -14 to -16 LUFS; raw phone recordings often sit 10 or more dB below that.
• Clipping. Recording too hot distorts transients permanently. Unlike the other three, no tool can fully undo it.

Notice what is not on the list: an expensive microphone. A $30 wired lavalier held close in a quiet room beats a $400 condenser two meters away in a reverberant one, every single time.

What to do about it

First, fix what recording technique fixes for free: get the mic closer, close the window, record a few seconds of silence so you know your noise floor. That is half the battle and costs nothing.

The other half happens after recording: reducing the noise that got in anyway, taming the room, and bringing the level up to what platforms expect without letting the peaks distort. That is mastering, and it is the half we built FixAudio for. But even if you never use our tool or any tool, move audio up your priority list. The research says your viewers already did.