In this recent test we put the three popular AI solutions against one another in a blind test; Adobe Podcast, Hush, and Descript . Here are the results.

The Three Contenders

Descript Studio Sound

Part of the Descript transcription solution which features Studio Sound for cleaning audio. Descript has a free option as well as paid plans. To use Studio Sound you need to upload audio to the cloud.

Adobe Podcast

Currently in beta, Adobe Podcast offers a similar record, transcribe and audio cleaning set of tools. No pricing has been announced to date. To use Adobe Podcast you need to upload audio to the cloud.

Hush

The newest in the pack, Hush is a machine based macOS application, with one clear aim, to clean up audio. Pricing is £49.00. Hush doesn’t upload to the cloud but works locally.

One point to make about all these apps. They do not have professional workflows, although Hush are considering both an AU and AudioSuite option.

Time Taken To Process The Audio

Below is the time taken to process each piece of audio. All of them are processing audio in less than a minute, however it’s worth noting that if you use these tools a lot then the difference between 30 seconds and 2 seconds is soon going to mount up and add a lot of time to your working day. Also, Hush can process mulitple audio clips at the same time, this is not presently the case for Descript and Adobe Podcast.

The Audio Examples

Here are the audio examples used. All audio has been loudness matched.

The Results

Summary

As we can see, there is no clear winner in the entire test. Each application won in a category, although when taking into account speed and privacy, Hush is an option worth considering as a professional workflow option.

Given that Adobe is hoping to win over podcasters, the poor result given in the narration test does mean they have some work to do on their algorithm.

Descript Studio Sound performed best across the board, gaining either first or second place in all tests.

As it’s early days for AI based noise reduction, and we expect algorithms to improve, these tests show encouraging results.

Hush are now working on both an AudioSuite AAX version and improving their algorithm, so watch this space!