Adobe is using AI to break down cluttered audio

Adobe is developing a new audio tool designed to break down the different sound layers within a single recording. The tool, called Project Sound Lift, uses AI to separate elements like applause from someone’s voice.

As shown in the demo sent to edge, all you have to do is import an audio file into the app and select the sounds you want the tool to filter out. There are many options to choose from, including applause, laughter, sirens, speeches, crowds, traffic. Project Sound Lift will automatically detect each sound and produce a separate file containing background noise and the tracks you want to prioritize, such as a The sound of a human voice or an instrument.

You can then import and edit each track individually in AdAdobe Premiere Pro, adjusting the volume of background noise while enhancing the clarity of the main track. It’s still early days, though—Adobe’s Project Sound Lift is being billed as a “sneak peek” preview from Adobe, which means we’ll have to wait a while to see the final version of the tool released (if Adobe if you decide to do so).

There are some other tools with similar functionality such as Ultrasonic clarity and Goyo sound separator, but they only offer options to isolate and adjust the file’s ambience, voice, and voice reverb. Adobe’s Project Sound Lift, on the other hand, is tailored to pick up specific sounds.

This is the same as for restore and release The Beatles’ first song in nearly 30 years: “Now and Then”. The producer was tasked with recovering John Lennon’s voice from the cassette tapes and using artificial intelligence to separate the overwhelming piano music from Lennon’s voice.