How to Remove Background Noise From a Podcast

Updated 22 August 2026

Almost every podcast recorded outside a treated room picks up something it should not: an air conditioner, a computer fan, traffic, the low hum of the building. Most of that is removable. Some of it is not, and knowing which is which saves you from destroying a recording trying to fix it.

What denoising can and cannot remove

Noise reduction works by identifying sound that is constant and predictable, then attenuating it. That distinction decides everything about what you can rescue.

  • Removable: air conditioning, fan noise, electrical hum, tape hiss, steady traffic, building rumble
  • Not removable: a door slamming, a dog barking, a chair scraping, someone talking in the background
  • Partly removable: intermittent noise that overlaps speech, at the cost of damaging the speech

The order to process in

Sequence matters more than people expect. Normalising a noisy recording raises the noise along with the voice, and denoising afterwards then has a louder problem to solve. Reducing noise first and setting levels afterwards consistently produces a cleaner result.

  • Denoise first, while the noise floor is as low as it is going to be
  • Then remove long silences to tighten pacing
  • Then normalise so episodes play at a consistent level

The mistake that makes voices sound underwater

Almost everyone over-processes on their first attempt. Speech occupies the same frequency ranges as much of the noise, so aggressive attenuation removes parts of the voice too. The result is that thin, watery, slightly metallic quality that listeners describe as sounding "processed". A recording with a faint audible room tone sounds more professional than one scrubbed until the voice warbles — leave some noise in.

What actually fixes this permanently

Post-processing cannot recover detail that was never captured. Moving the microphone closer to the speaker raises the voice relative to the room, which improves the ratio at the source rather than trying to repair it afterwards. Turning off the air conditioning for the recording, and putting something soft on hard surfaces, does more than any filter. Thirty seconds of setup beats an hour of cleanup.

Why do this locally

Podcast audio is unreleased content, frequently including guest conversations that were recorded under an expectation of privacy until the episode ships. Uploading raw recordings to a cloud audio service to clean them is a bigger exposure than most people register. Processing in the browser means the raw recording never leaves your machine.

The Audio Denoiser runs entirely on your device — nothing is uploaded.

Open the Audio Denoiser →

Read more about how it works

Frequently asked questions

Can I remove someone talking in the background?

Not really. Denoising targets steady, predictable sound. Another voice is neither, so it passes through the filter largely intact.

Why does my voice sound strange after noise removal?

The filter is set too aggressively. Speech shares frequencies with the noise, so heavy attenuation strips part of the voice. Back it off until the voice sounds natural, even if some room tone remains.

Should I normalise before or after denoising?

After. Normalising first raises the noise floor along with the voice and makes the noise harder to remove cleanly.

Is it worth re-recording instead?

If the noise is intermittent — barking, banging, background conversation — often yes. No filter removes those cleanly, and a re-record is usually faster than fighting it.

Related guides