Noise Reducer
Clean up hiss, hum, fans, and room noise in voice memos, podcasts, and vocal demos. The reducer finds a quiet moment, learns what the noise sounds like, and removes it from the whole recording, right in your browser.
Works best when the recording includes a second or two of silence, like the pause before you start talking.
How to remove background noise from audio
- Drop in a recording (MP3, WAV, M4A, FLAC, OGG, or AAC).
- Pick a strength, and turn on hum removal if you hear an electrical hum.
- Press Reduce noise. The tool learns the noise from a quiet gap and cleans the whole file.
- Listen to the result, then download it as MP3 or WAV.
How it works
Steady background noise has a fingerprint: the hiss of a preamp, the whoosh of an air conditioner, the drone of a laptop fan. The reducer scans your recording for the quietest stretch that still has sound in it, usually the pause before you start talking, and learns that fingerprint. Then it runs spectral noise reduction over the entire file, pulling down whatever matches the fingerprint while leaving your voice or instrument alone.
If the recording has no quiet gap at all, it switches to automatic mode, which estimates the noise as it goes. That still helps, but more gently, so you will get the cleanest result from recordings that include a second or two of room tone.
What it can and cannot remove
It is very good at steady noise: hiss, hum, buzz, fans, air conditioning, computer noise, and constant traffic rumble. It cannot remove noise that changes from moment to moment, like a dog barking, people talking nearby, music playing in the room, or keyboard clicks, because those sounds do not match a single fingerprint. Heavy echo from a bare room is a different problem too, and noise reduction does not fix it.
For electrical hum, turn on hum removal and pick your mains frequency: 60 Hz in North America, 50 Hz in Europe, the UK, Australia, and most of Asia and Africa. It cuts the hum and its next three harmonics with very narrow notches, so the rest of the sound is untouched.
Getting a natural result
Start with Medium. Gentle keeps the recording most natural and suits noise you can barely hear. Strong removes more, but pushing noise reduction hard on a noisy recording can leave a watery, underwater quality on the voice, so save it for when the noise is the bigger problem. Clean the noise before you add compression or boost the volume, because both make the noise louder along with everything else. And the best noise reduction still happens at the source: get closer to the microphone, switch off the fan, and record a few seconds of silence first.
Noise reducer: common questions
How do I remove background noise from a recording?
Drop the file on this page, leave strength on Medium, and press Reduce noise. The tool finds a quiet gap, learns the noise, and cleans the whole file on your device. Listen to the preview, try Gentle or Strong if needed, then download.
Why does my voice sound watery after noise reduction?
That is the classic sign of too much reduction. When the noise is loud compared to the voice, removing all of it takes some of the voice with it. Switch to Gentle, or accept a little background noise, which usually sounds more natural than an over processed voice.
Can it remove wind, music, or people talking?
Not well. The reducer removes steady noise that sounds the same from moment to moment. Wind gusts, voices, barking, and music change constantly, so they do not match the learned noise. For music behind a voice, a stem splitter that separates vocals, like the one in Xport Studio, is the better tool.
How do I remove hum from audio?
Set hum removal to your mains frequency: 60 Hz in North America, 50 Hz in Europe, the UK, Australia, and most of Asia and Africa. The tool cuts that frequency and its next three harmonics with very narrow notches. If you are not sure which you have, try one, listen, then try the other.
Can it remove static or hiss from old recordings?
Yes, steady hiss and constant static are exactly the kind of noise it handles best, including tape hiss. Crackles and pops are different: each click is a separate sudden sound rather than a steady one, so they mostly remain.
Is this an AI noise remover?
No. It uses classic spectral noise reduction from FFmpeg rather than a neural network. That keeps it fast, private, and free of big downloads, and it works best on steady noise. AI speech enhancers can cope better with changing noise, but most of them need you to upload your recording.
Is my recording uploaded?
No. The noise reducer is FFmpeg compiled to WebAssembly, running inside your browser tab. Your recording is processed on your own device and never leaves it, so we never see, store, or train on it.
Who makes this?
We're the small team behind Xport Studio, a private, fully offline app for AI vocals, stems, and mastering. These free tools run on the same idea: your audio stays on your device. If you're curious about the bigger thing, take a look.