Free tools / Gentle clean-up

a light touch

Gentle clean-up

Take out low rumble and mains hum, and lightly lower steady background noise like a fan or fridge. Listen to the before and after, then save a new WAV. It's deliberately gentle, so your voice still sounds like you.

Your file is read and processed on your device. It isn't uploaded anywhere.

What to clean

Hum comes from the mains electricity, so it's at 50 Hz or 60 Hz depending on where you are, plus a few multiples of it.

6 dB

0 turns it off. Higher settings can start to make a voice sound watery, so go as low as you can.

before and after

Choose a file, pick what to clean, then press Clean it.

What it does, and what it doesn't

  • Low rumble. A filter takes out sound below the frequency you choose, at 24 dB per octave. Most of a speaking voice sits above that, but a very deep voice can reach down near 100 Hz, so start at 80 Hz and listen.
  • Mains hum. Very narrow cuts at 50 Hz or 60 Hz and the next three multiples (for example 100, 150 and 200 Hz). Each cut is narrow enough to leave the rest of the sound alone.
  • Background noise. It learns the sound of your room from the quietest half-second in the file (shown on the waveform), then turns down that steady noise where your voice isn't covering it. The reduction is capped at the level you set, up to 12 dB, so it stays gentle.

It can't remove echo, a room that sounds boomy, voices, music or sudden noises like a dog barking. If those are on a take, the best fix is usually to record it again somewhere quieter and softer. The room noise check helps you find that spot.

Keep it natural

Voicefolk's rules ask for natural audio: no heavy processing, pitch-shifting or voice changing. This tool doesn't change pitch or the character of your voice, and it doesn't use AI. Listen to both versions before you send anything, and if the cleaned one sounds less like you, use less reduction or the original.

Everything runs in your browser. Your recording isn't sent anywhere.

Then check the level

Once it's clean, the loudness normaliser sets the level the client asked for, and the audio file checker checks it against their spec.