Start by identifying the noise, not by reaching for a preset

A noisy webinar recording is usually a mix of problems: a laptop fan, a guest on a headset mic in a tiled room, keyboard clicks, notification dings, and a low electrical hum. Noise removal works best when you match the tool to the noise. Listen on headphones, write down timestamps, and solo each track if you have separate files. If the noise changes between host and guest, one setting will not fix the whole timeline.

Common webinar noises and their fixes

If the audio is clipped — flat-topped waveforms, crackling on loud laughs — repair that before noise removal. If the room sounds like a bathroom, de-reverb matters more than a noise gate.

Prepare the file so cleanup does not create new artifacts

  1. Duplicate the original. Never process the only copy. Keep a clean master in a separate folder.
  2. Work in WAV, 48 kHz, 24-bit. MP3 and other lossy formats make noise removal harder because codec artifacts get amplified. Export to MP3 only at the end, at 192 kbps or higher.
  3. Isolate the noisy track. In a multitrack webinar, process only the participant with the problem. Do not apply the same noise reduction to a clean host track.
  4. Split the timeline when the noise changes. If the first 20 minutes have air conditioning and the rest is quiet, use different settings for each section.

For a single mixed track, you will have to compromise. A one-click tool can save time, but check the result on headphones and a phone speaker.

The cleanup chain that actually works

Order matters: corrective cleanup first, then tonal shaping, then dynamics. If you compress before removing noise, the compressor will bring up the noise floor along with the voice.

1. High-pass filter to remove rumble

Set the high-pass filter at 80 Hz for a deep male voice, 100–120 Hz for a higher voice. Use a 12 dB per octave slope. If the speaker sounds thin, lower it to 60–70 Hz. Do not use a high-pass filter as your only noise removal. It will not touch hiss or keyboard clicks.

2. Broadband noise reduction for hiss and hum

Find one or two seconds where only the noise is playing. Learn that noise profile, then reduce by 6–12 dB. Listen for watery, metallic, or robotic artifacts. If you hear them, back off or use a gentler setting. For hum, a notch filter at 50 Hz or 60 Hz plus harmonics can be cleaner than broadband reduction.

If you do not want to learn a DAW, a free online option like ClearVoiceHQ (clearvoicehq.com) can remove background hiss, hum, and room noise from remote meeting, webinar, and interview recordings. Upload the file, choose a reduction strength, and start with medium rather than aggressive. A 6–10 dB reduction that sounds natural beats a 25 dB reduction that makes the speaker sound like a robot.

3. Noise gate or expander for intermittent noise

A gate mutes the signal when it drops below a threshold. Set the threshold just above the noise floor, often between -45 dB and -35 dB for a quiet room, or -30 dB for a noisy one. Use a range of -12 to -24 dB, not -80 dB. Attack 5–10 ms, hold 100–200 ms, release 150–300 ms. If the gate chops breaths, use an expander instead. Never gate a track with two open mics; the other speaker will trigger the gate and cause pumping.

4. EQ for intelligibility

Cut mud between 200 Hz and 400 Hz by 2–4 dB if the voice sounds boxy. Boost presence between 2 kHz and 5 kHz by 1–3 dB if the voice sounds dull. If sibilance is harsh, use a de-esser at 5–8 kHz. Do not boost high frequencies to mask hiss; that only makes the hiss louder. Sweep a narrow boost to find a problem frequency, then cut it.

5. De-reverb for room echo

De-reverb is not the same as noise reduction. Use it sparingly, typically reducing 20–40 percent. Push it harder and the voice can sound metallic or hollow. For severe echo, rerecord the intro and outro, or add captions. A little room sound is better than a processed, unnatural voice.

6. Compression and loudness

After cleanup, compress with a ratio of 2:1 or 3:1, threshold around -18 to -12 dB, attack 10–30 ms, and release 100–200 ms. Aim for -16 LUFS integrated for web video, or -16 to -19 LUFS for spoken-word audio. Set true peak at -1.5 dB. If the noise floor rises too much under compression, remove more noise before compressing.

Example settings for a typical Zoom webinar

Imagine a host on a USB mic and a guest on a laptop mic in a small office. The guest has fan noise and a boxy tone.

These are starting points, not laws. Bypass the chain often and compare at matched loudness.

Mistakes that make noisy webinar audio worse

When to use one-click noise removal versus manual editing

If you have a single mixed track, a deadline, and mostly steady hiss or hum, a one-click tool like ClearVoice is often enough. If you have separate tracks, severe echo, or intermittent keyboard noise, manual editing in a DAW gives you more control. You can also combine them: run the online cleaner first to lower the broadband noise floor, then open the cleaned WAV in your editor for EQ, de-essing, and compression.

Bottom line

Clean up noisy webinar recording audio by diagnosing the noise, preparing a lossless copy, and applying a light-touch chain: high-pass, noise reduction, gentle gating, EQ, de-reverb, and compression. Do not chase silence. Chase clarity. A one-click online tool can handle broadband noise quickly, while a DAW gives you control to fix specific problems. Before export, A/B at matched loudness and check on a phone speaker. Work on a copy, listen on real-world speakers, and stop when the voice sounds natural.