Free · private · in your browser
Reduce fans, hiss and room noise while keeping speech clear.
Air conditioning, a laptop fan, traffic, mains hum and microphone hiss all sit under a recording as a steady bed of noise. This tool reduces that constant background while leaving speech intelligible, which is the difference between a video that sounds amateur and one that sounds fine — and it is far cheaper than re-recording.
Noise reduction works by learning the profile of a constant sound and subtracting it. That is why it handles fans, hum, hiss and distant traffic well: they do not change much over time. It cannot do anything about a door slamming, a chair scraping, a dog barking or a siren passing, because those are one-off events with no stable profile. Sudden noises have to be cut out or covered, which is a job for section removal or a music bed rather than for a denoiser.
Push the reduction too far and speech starts to sound like it is coming through a tunnel, with a watery, warbling quality around consonants — the classic artefact of over-processing. That is because the noise you are subtracting overlaps in frequency with the voice you are keeping, so past a point you are removing both. A moderate setting that leaves a little audible background is almost always more pleasant than an aggressive one that leaves speech sounding synthetic. Compare against the original rather than judging the processed version alone.
Steady, continuous sounds: fans, air conditioning, mains hum, tape and mic hiss, and constant distant traffic. Those have a stable profile that can be learned and subtracted.
No. One-off sounds have no steady profile to subtract. Cut those moments with the section remover, or cover them with music.
The reduction is set too high. Noise overlaps the voice in frequency, so aggressive settings remove parts of the speech too, which produces the watery, tunnel-like artefact. Back it off.
Before. Noise reduction works best on the original recording, and later steps such as normalising or compressing would otherwise amplify the background along with the voice.
No. It reduces a constant background but cannot recover detail the microphone never captured, undo clipping, or remove echo from a reflective room.
