Clean Audio for removing background noise from recordings

Clean Audio reduces background noise in spoken audio and video recordings. Its product page describes a speech-trained model that runs on the user's device in the default browser mode. This is useful for checking an interview, podcast clip or spoken video before deciding whether it needs more editing.

Cleaning and comparing a recording

Load a supported file and compare the Original and Cleaned versions. The publisher lists audio inputs such as MP3, WAV, M4A and FLAC, as well as common video containers. A cleaned audio recording can be exported as MP3 or WAV. The page says cleaned video generally returns as MP4, or as WebM for a WebM input.

The noise reduction model is trained for speech. The publisher describes it as separating voices from changing or steady background sounds. That is a reason to listen for loss of sounds you intended to keep, particularly if the recording contains music or effects as well as a speaker.

Local and cloud processing

The default tool downloads the noise model to the browser and processes the recording on the device. The publisher lists local limits of 30 minutes and 2 GB, and recommends a computer for longer files. It describes Studio as an optional cloud mode that uses a larger model. Choosing Studio sends the audio to the cloud for that job. The publisher says that upload is deleted within a day.

Cost and when to use Studio

The browser tools are free. The Studio plan charges by the minute beyond an allowance, and the pricing page says an account gets two free Studio minutes per day. Try the browser version and compare the same section in both playback modes before choosing Studio. If you cannot send the recording off-device, keep it in the local browser mode.