Free tool

Remove background noise from audio and video.

Take the fan, the hum, the hiss, the traffic and the keyboard out of a voice recording or a screen recording and keep the voice. Drop a sound file or a video; a video keeps its picture exactly as it was. An open AI model runs in this tab, so nothing is uploaded and there is no sign-up.

Drop a recording here, or click to choose MP3, WAV, M4A, OGG or FLAC, or a MOV, MP4 or WebM video. It stays on your computer.

This runs in your browser. Your recordings are never uploaded.

How it works

Three steps, nothing to install.

01 / Drop

Choose a recording

A voice memo, a podcast, an interview or a screen recording with your voice.

02 / Enhance

Let the AI remove the noise

The first time, the page downloads the AI model, 34 MB. Then your computer cleans the sound, much faster than real time.

03 / Compare

Listen and download

Switch between the original and the enhanced sound at the same moment, then save it.

Questions

Good to know.

Is my recording uploaded?

No. The AI model is downloaded to your browser and runs there, so the recording never leaves your computer. There is no account, no length limit and no watermark.

Which AI model does it use?

DeepFilterNet3, an open speech enhancement model by Hendrik Schröter and others, released under the MIT and Apache 2.0 licences. It was trained to keep speech and remove everything else, at 48 kHz, the quality of a video's sound.

How much noise does it remove?

In our test, a spoken clip with loud fan-like noise and a 50 Hz hum mixed in, the noise in the pauses went from -26 dB to -62 dB, while the voice stayed at the same level. Real rooms differ, so listen to the before and after.

Does it remove echo?

A little, not fully. A big echoing room still sounds like a room. Medium or Light noise removal keep some of it on purpose, which can sound more natural.

Will it work on music?

No. It is made for speech and treats music as noise, so it will take most of the music out.

What happens to a video?

The picture is copied as it was, with no quality loss, and the sound is replaced with the clean version, in sync. MOV and MP4 come back as MP4, WebM as WebM. The new sound is mono on both speakers.