Why podcast background noise is stubborn

Fan hum, laptop whine, room tone, keyboard clatter, traffic, and electrical buzz are not the same problem. Noise removal separates wanted voice from unwanted sound. Steady noise like a fan or AC is easier because the denoiser learns its shape from a quiet moment. Intermittent noises like a dog bark, mouse click, or chair creak need manual or spectral repair. Too much processing makes voices sound underwater, metallic, or lispy. Your goal is not perfect silence; it is a lower noise floor while keeping the voice natural.

Before you touch a plugin: 4 fixes that save time

  1. Record 3-5 seconds of room tone at the start and end of every session. It is your noise print. If you forgot, use the longest pause between sentences.
  2. Back up the raw file and work on a copy. Noise removal is destructive in many editors.
  3. Check mic technique: 4-6 inches from your mouth, slightly off-axis, peaks around -12 to -6 dB. If you recorded too hot, re-record instead of fighting noise.
  4. Separate tracks if you can. In OBS, mic to track 1 and game/system to track 2. In Zoom, enable local recording for separate WAV tracks. Process only the voice track.

Step-by-step noise removal workflow

1. High-pass filter first

Cut rumble before denoising. Male voices: start at 80-100 Hz. Female voices: 100-120 Hz. Use a 12-24 dB per octave slope. In Audacity, use Effect > Filter Curve EQ or High-Pass Filter. In Premiere Pro, use Parametric EQ with a high-pass band. Do not cut up to 150 Hz or the voice will sound thin.

2. Remove electrical hum

If you hear a steady low buzz, it may be 60 Hz in the US or 50 Hz in Europe, plus harmonics. Notch the fundamental and harmonics at 120/180 Hz or 100/200 Hz. Use iZotope RX De-hum, Audacity Notch Filter, or ReaFIR. Keep the notch narrow; over-notching can hollow out male voices.

3. Noise reduction or denoise

Select 1-2 seconds of room tone. In Audacity, go to Effect > Noise Reduction > Get Noise Profile. Select the whole track and open Noise Reduction again. Start with Reduction 6-12 dB, Sensitivity 6-8, Frequency smoothing 3. Preview. If the voice sounds robotic, lower the reduction. For nasty noise, do two passes at 6 dB instead of one at 18 dB. In Adobe Audition, use Noise Reduction or DeNoise. In iZotope RX, use Voice De-noise at 6-10 dB. If you want a free browser option for steady fan, AC, or hiss, ClearVoice can clean a file without a DAW. Upload, run noise removal, download, then finish with EQ and gating.

4. Gate or expander

After denoising, you may still hear a low bed between phrases. Use an expander instead of a hard gate. Start with threshold 3-6 dB above the noise floor, ratio 2:1 to 4:1, attack 5-10 ms, release 100-200 ms. If words get chopped or breaths disappear, lower the threshold or ratio. In OBS, use Noise Gate. In Premiere Pro, use Dynamics > Expander.

5. Fix clicks, plosives, and sibilance

Noise removal will not fix a keyboard click or mouth smack. For single clicks, use Audacity Repair or iZotope RX De-click. For plosives, a high-pass filter helps; if a P or B pops, fade that syllable or cut 100-150 Hz with narrow EQ. For harsh S sounds, de-ess around 5-8 kHz. Be gentle; over-de-essing makes speech lisp.

Settings by source: podcast, Zoom, phone, gaming

Podcast mic recording

Best case: isolated track. Use high-pass, de-hum if needed, denoise 6-10 dB, expander, then EQ. Aim for a noise floor below -60 dB in silence. Record 24-bit WAV at 48 kHz if possible. Do not use MP3 until final export.

Zoom call

If local recording is on, you get separate WAV tracks. If you only have cloud recording, you get a mixed M4A. For mixed tracks, use light denoise at 4-8 dB. Ask everyone to mute when not speaking so gaps are clean. If Zoom noise suppression was already on, do not stack heavy denoise. For a mixed Zoom file, a free tool like ClearVoiceHQ (clearvoicehq.com) can remove steady room noise, but you still need to cut cross-talk and long pauses.

Phone voice memo

iPhone Voice Memos record AAC, which is fine for editing only if you convert to WAV first. Hold the phone 4-6 inches below your mouth, not directly in front. Record in a closet or under a blanket if the room is echoey. In post, high-pass at 100 Hz and denoise 8-12 dB. Phone audio often sounds boxy around 200-400 Hz; cut 2-3 dB there if needed. Do not add bass to compensate.

Gaming clip or streaming VOD

If you recorded with OBS, keep mic and game audio on separate tracks and process only the mic. If the clip is already mixed, noise removal will also affect game audio, so use it lightly. For keyboard noise, a gate or expander helps, but a dynamic mic close to your mouth helps more. For live use, NVIDIA Broadcast or RTX Voice can reduce keyboard and fan noise before recording. For recorded clips, use spectral repair for isolated clicks and leave game audio alone.

Free and paid tools: when to use which

Audacity is free and has solid noise reduction, but spectral repair is clunky. Adobe Podcast Enhance is free on the web and good for speech, though it can over-smooth. iZotope RX is paid and best for salvage. Descript Studio Sound is paid and easy. Premiere Pro, DaVinci Resolve, and Final Cut Pro have built-in cleanup. If you need to remove one dog bark or a single keyboard click, use spectral repair.

Common mistakes that make podcast audio worse

Quick checklist before export

  1. Listen on headphones and a phone speaker.
  2. Check silence: aim for a noise floor below -60 dB.
  3. Check S sounds and breaths for metallic artifacts.
  4. Bypass the processing and compare. If processed sounds worse, lower the settings.
  5. Export WAV at 48 kHz, 24-bit. Then make an MP3 at 192-320 kbps for publishing.

Background noise removal is a chain, not a single button. Get the cleanest recording you can, capture room tone, denoise gently, gate lightly, and fix clicks by hand. That workflow works for a podcast, Zoom call, phone memo, gaming clip, or YouTube video, and it keeps your voice sounding like you.