Remove background noise from audio, free and online
Upload a recording with noise under the voice: a fan, traffic, a busy café, a humming fridge. The AI keeps the speech, removes what is around it, and brings the voice up to a clear, even level. You hear it against the original before you download.
No sign-up for 3 files (6 min) · deleted after 7 days · never used for training
Drop a noisy recording here
Audio files, up to 3 minutes
AAC · FLAC · M4A · MP3 · OGG · OPUS · WAV
Free: 3 files up to 3 minutes each (6 minutes in total), no account.
Before and after
Only the voice is left, at the same level all the way through.
How to remove background noise from a recording
- Step 1
Upload the file
Choose an audio file (video files need Premium). It goes straight to processing; there is nothing to set up first.
- Step 2
Compare before and after
Play the result and switch between it and your original while it plays. The playhead stays where it is, so you compare the same words.
- Step 3
Download
Set the output volume if you want it louder or quieter, then download MP3 or WAV.
What it removes
Steady noise is the easiest to hear and the easiest to remove: fans, air conditioning, computer hum, the hiss of a microphone turned up high. Changing noise is harder for simple filters but handled the same way here: passing traffic, people talking in the background, keyboard clicks, dishes, wind.
The model is trained on speech, so it does not try to guess what the noise is. It recognises the voice and treats everything else as something to remove. That is why it also takes out room echo in the same pass.
What it cannot do
It is built for a speaking voice. Music playing under the voice is better handled by Separate music, which splits a recording into the voice and the music. A second person talking over the first counts as speech and stays. And if the noise was so loud that a word cannot be heard at all, that word is not in the file to recover.
Tips for a cleaner result
- Upload the original file, not one that has already been through a noise gate or heavy compression.
- Get the microphone close to the speaker when you record; distance adds both noise and echo.
- If you recorded several people on separate tracks, clean each track on its own.
Questions about removing background noise
Will removing the noise hurt my voice?
The model separates the voice from the noise and keeps the voice, rather than cutting frequencies the way an equaliser does. You can hear the result against your original before you download, and keep the original if you prefer it.
What kinds of noise does it remove?
Steady noise such as fans, air conditioning, hum and hiss, and changing noise such as traffic, chatter, keyboard clicks and wind. Room echo is reduced in the same pass.
Does it work if there is music behind the voice?
Use Separate music for that. It splits a recording into the voice and the music, and you can download the voice on its own.
Is it free?
Yes. Without an account you get 3 files up to 3 minutes each, 6 minutes in total, with no sign-up. A free account raises that to 30 minutes per file and 60 minutes of audio per day, still with no card. Premium allows up to 5 hours per file, 10 hours of audio per day, and fair use of 100 hours per 30 days.
Which files can I upload, and what do I get back?
On every plan: audio files (WAV, MP3, M4A, AAC, FLAC, OGG and Opus). Video files (MP4, MOV, WebM, MKV, AVI and WMV) need Premium; the sound is taken from the video and you download the cleaned audio. Results download as MP3 or WAV.
What happens to my recording?
It is used only to make your result. It is never used to train models, and both the upload and the result are deleted automatically after 7 days.