AI audio cleaner for voice recordings
An AI model trained on speech separates the voice from the noise around it. It runs in your browser, so the recording stays on your device.
- Choose a file
- Press "Remove noise"
- Download the WAV
For speech: voice notes, interviews, podcasts, classes. You download the clean audio as a WAV, which plays on any phone or player.
Hear it on real recordings
Voice recording in a kitchen Removed: kitchen noise, dishes, fridge hum
Voice message Removed: background noise around the voice
Your recording never leaves your device
How to do it
- Choose a recording with speech: voice note, interview, podcast, class or call.
- Press "Remove noise" and wait for the bar.
- Compare before and after, then download the WAV.
Tips
- It cleans noise around a voice; it does not change the voice itself or add effects.
- Best on steady noise: fans, hum, traffic, a café. Very loud sudden sounds may stay.
Questions
Which AI does it use?
DeepFilterNet3, an open-source noise filter trained on speech (MIT / Apache 2.0). It is downloaded once, about 12 MB, and runs on your device.
Is it really free?
Yes: no sign-up, no length limit and no watermark.
Does it enhance music or singing?
No. The filter keeps voices and treats everything else as noise, so it damages music.