British Columbia-based independent programmer Ian Sampson has announced the launch of Hush, a new macOS app that uses artificial intelligence (AI) to remove background noise and reverb from spoken audio.
Hush builds on transformer architecture, the technology that powers generative models like ChatGPT and DALL·E 2, with the goal of isolating speech with greater precision and fewer artifacts than previous approaches. It is designed to suppress broadband noise, transient sounds like barking dogs or honking horns, and reflections from untreated rooms.
Hush runs on the Neural Engine, a subprocessor for machine learning on Apple Silicon Macs, which allows Hush to render audio more quickly and efficiently without heating up the computer or spinning up the fans.
The new app supports macOS 12 (Monterey) or macOS 13 (Ventura), with an M1 or M2 Mac is recommended for optimal performance. The app does not require iLok or other third-party copy protection.
Hush is available now on the Mac App Store for an introductory price of $49.99 (U.S.), with the app as well as sample audio and a 21-day free trial available here.
Developer Ian Sampson has published several open-source libraries on GitHub, with Hush serving as his first commercial release. Reach him at [email protected].