A voice memo recorded next to a fan, a lecture captured from the back of a hall, a podcast with traffic behind it: most recordings pick up steady background noise, and the usual fix is an upload to a server, an account, or a paid plan. This free remove background noise from audio tool cleans the recording in your browser instead. DeepFilterNet, an open source AI model built for speech, lowers the fan hum, hiss and room tone while keeping your voice natural, and the cleaned recording downloads as a WAV. It is free, needs no sign-up, and runs entirely on your own device, so your recordings are never uploaded to a server.
Unlike the upload-and-wait cleaners that cap free use at a few minutes a day, everything here happens locally, with no daily limit and no account. You can also drop in a video file: the tool reads the audio track from an MP4, WebM or MOV file, cleans it, and gives you back the cleaned audio as a WAV. There is no per-minute fee and no file count limit, only the size of the file itself.
Key features
- Real noise removal: DeepFilterNet3 lowers steady background noise such as fan hum, air conditioning, hiss, traffic drone and room tone, while keeping speech intact. In our test on a real voice recording with added fan and hum noise, the background noise fell by about 13 dB and the speech became about 6 dB clearer against the clean original.
- Strength and room tone controls: strength sets how far the noise floor is allowed to fall, and room tone mixes a little of the original recording back in, so the voice does not sound over-processed or cut off between words.
- Audio and video files: MP3, WAV, M4A, OGG, FLAC, MP4, MOV and WebM, up to 100 MB and 20 minutes. Video files work when your browser can decode their audio track.
- Compare before and after: listen to the original and cleaned recordings side by side, with a noise reading for each, before you download.
- Private by design: the recording is decoded and cleaned on your device. There is no upload, no account and no copy left on a server.
How to remove background noise from audio
Drop your audio or video file onto the tool, or choose it from your device. Pick a strength, then click Clean this recording. The first run downloads the noise removal model once, about 17 MB, and your browser then caches it. Cleaning runs on your device and takes a while on long files: a 15 second clip takes seconds, a 10 minute recording takes a few minutes. When it finishes, compare the original and cleaned recordings, check the noise reading, then download the cleaned WAV.
Honest limits
This tool reduces steady background noise. It does not remove background music, because music is not steady noise, and a loud music track can confuse the cleaner. It does not fully remove echo or reverb, because those are the voice itself bouncing back, not background noise. It cannot rebuild speech that was never captured clearly: a voice buried under much louder noise, clipped audio, or overlapping speakers will still sound rough after cleaning. The cleaned file is a fresh WAV at 48 kHz, so it is not bit-identical to the original, and files are capped at 100 MB and 20 minutes because all the work happens on your device.
Private by design
Like every Savozi tool, this one runs entirely in your browser. Your recordings are never uploaded to our server, so there is no server copy and no account holding your audio. It is part of our watermark tools collection: clean up photos and videos, trim clips, save video audio as MP3, transcribe recordings to text, and now clean the background noise from a voice recording, all free, with no account and nothing to install.