Skip to main content

audio-noise-reducer

An audio noise reducer suppresses background noise — fan hum, computer whirr, traffic, hiss, room reverb — from a recording, producing a cleaner voice or music track suitable for podcasts, video voice-overs, and shareable recordings. The ZTools Audio Noise Reducer runs entirely in the browser using a Web Audio API spectral noise gate, lets you sample a "noise-only" segment of the recording so the algorithm knows what to remove, supports adjustable strength (light to aggressive), and exports the cleaned audio as MP3 or WAV — without uploading the source file.

Use cases​

Podcast / interview cleanup​

Recorded in a room with HVAC hum or laptop fan noise. The tool isolates and removes the constant background, leaving voice intact. Listeners notice less fatigue over a 30-minute episode.

Field-recording cleanup​

Outdoor recording with wind / traffic. Spectral noise reduction can recover some intelligibility lost to ambient noise (within limits — heavy noise destroys the underlying audio).

Voice-memo polish​

A voice memo recorded in a coffee shop. Light noise reduction removes the espresso machine without making the voice sound hollow.

Pre-transcription cleanup​

Speech-to-text engines accuracy drops sharply with background noise. A pre-transcription noise reduction can lift accuracy from 70% to 90%+.

How it works​

  1. Upload audio — MP3 / WAV / OGG / M4A. Tool decodes to PCM in browser memory.
  2. Identify noise sample — Mark a 1-3 second region containing only background noise (no voice / music). Tool builds a spectral profile of "what to remove".
  3. Apply noise gate — For each FFT frame of the recording, frequencies matching the noise profile are attenuated below a threshold. Threshold is tunable (light / medium / aggressive).
  4. Preview — Listen to the result. Aggressive settings can introduce a "watery" artifact (musical noise); back off to medium if too obvious.
  5. Export — Download cleaned audio as MP3 or WAV. Original untouched.

Examples​

Input: 5-min podcast with HVAC hum, sample 2 sec of silence

Output: Hum removed; voice intact. Light artifact only on aggressive setting.

Input: Interview with laptop fan whirr

Output: Fan removed; voice clarity improves measurably; transcription accuracy up.

Input: Outdoor recording with wind gusts

Output: Constant wind component reduced; gust transients harder to remove. Use a wind muff in field for best results.

Frequently asked questions​

Why do I need to sample noise?

The algorithm needs to know what noise looks like spectrally before it can suppress it. Without a sample, it would also attenuate desired audio (music, voice) at the same frequencies.

What is "musical noise"?

An artifact of aggressive spectral noise reduction — residual noise becomes pitched bursts that sound watery / metallic. Caused by over-suppressing frequencies. Light settings avoid it; aggressive risks it.

Can I remove voices in the background?

No — voice has a complex, time-varying spectrum, similar to the foreground voice. The algorithm cannot separate two human voices reliably; that requires AI source separation (Demucs, SpleeT) which run on desktop.

How much noise can be removed?

Constant noise (hum, hiss, fan) ~10-25 dB suppression possible without obvious artifacts. Variable noise (people, music) much harder. Best result: avoid noisy recording in the first place.

Is the audio uploaded?

No — entirely client-side. Web Audio API processes in browser memory.

Why does my voice sound thin after reduction?

Aggressive setting attenuated voice frequencies that overlapped with noise frequencies. Reduce strength; or sample a quieter noise segment without voice resonance.

Tips​

  • Always record clean before relying on cleanup. Reduction recovers some quality but never restores what was masked.
  • Sample noise from a 2-3 second silent region with no voice — picking a region with breath or movement contaminates the profile.
  • Start with light reduction; bump up only if needed. Aggressive settings introduce more artifacts than mild noise.
  • For broadcast-quality results, follow noise reduction with EQ + compression to restore tonal balance.
  • Save a backup before processing — irreversible if you overwrite the source.

Try it now​

The full audio-noise-reducer runs in your browser at https://ztools.zaions.com/audio-noise-reducer — no signup, no upload, no data leaves your device.

Open the tool ↗

Last updated: 2026-05-05 · Author: Ahsan Mahmood · Edit this page on GitHub