WoluTools
Find a tool

Video, Audio & YouTube

Remove background noise and hum from voice recordings

Upload WAV, MP3, M4A, FLAC or MP4. Filters target hum, rumble and steady noise such as fans. You get the cleaned file plus before/after samples to compare.

or drop it here

WAV, MP3, M4A, FLAC, MP4 · Up to 15 minutes
Free account needed for your 3 free jobs a day

No file at hand? See the prepared example

Standard toolIncluded · 3 free jobs a dayFree: 3 jobs a dayPro: up to 200 jobs a day · €12.99/month or €89.99/year
  • WAV / MP3 / M4A / FLAC / MP4
  • Up to 15 minutes
Prepared result preview · fictional sample dataSynthetic voice with 50 Hz hum
Source
Apply a conservative high-pass, notch and FFT denoise chain
Measured interpretation
  • High-pass cutoff: 70 Hz
  • Confirmed mains hum: 50 Hz
  • Output: PCM WAV

cleaned.wav · manifest.json

BringWAV, MP3, M4A, FLAC, MP4
GetCleaned media, samples, spectra and manifest
PrivacyEncrypted source · 24-hour result

One clear job, from source to download

  1. 1

    Add the source

    Supported formats and limits are visible before the upload.

  2. 2

    Confirm the settings

    Review the exact source, options, units and access before processing.

  3. 3

    Inspect and download

    Check the preview and warnings, then unlock the complete package.

Reducing hum and steady noise in voice recordings

What you upload and what it can reduce

You upload one recording in WAV, MP3, M4A, FLAC or MP4, up to 15 minutes long. The filters target steady noise: mains hum, low rumble and constant background hiss from things like fans. Sounds that are not steady, such as a door slam, a cough or overlapping voices, stay in place. The speech itself is not regenerated or replaced, so words that were hard to understand in the source stay that way.

What you get back

You get a cleaned PCM WAV file, short sample clips and before and after spectra, plus a manifest. The manifest records the high-pass cutoff, the notch frequency and the denoise setting that were applied. Listen to the samples side by side and look at the spectra to see what changed. If the voice sounds thinner than before, run the job again with a lighter setting. Your original file is not changed.

Settings and the trade-off

Mains hum can be set to none, 50 Hz or 60 Hz. Pick the mains frequency of the place where the recording was made, 50 Hz in most of Europe and Asia and 60 Hz in North America. Denoise strength is light, medium or strong. Stronger settings remove more background but can thin consonants and leave artefacts around speech. The preview lets you judge that trade-off before you keep the result.

Questions before you run it

What kinds of noise can it actually reduce?

Steady noise: mains hum, low rumble and constant background hiss. A door slam, a cough or overlapping speech is not steady, so the filters leave it in place.

Should I pick 50 Hz or 60 Hz for mains hum?

Use the mains frequency of the place the recording was made, 50 Hz across most of Europe and Asia and 60 Hz in North America. The preview reports the hum frequency it confirmed, so the choice can be checked before rendering.

Can strong denoise damage the voice?

Yes. Heavier settings take out more background but begin to thin consonants and leave artefacts around speech. The preview and the before and after spectra exist so you can judge that trade-off.

Is the speech regenerated or replaced?

No. Only deterministic filters run over the recording you supply, and nothing is synthesised. Words that were unintelligible in the source stay unintelligible.

What comes back, and in which format?

A cleaned PCM WAV with sample clips and spectra, plus a manifest recording the high-pass cutoff, the notch frequency and the denoise setting that were applied.