At its annual Max conference, Adobe showed off a proof-of-concept project called Super Sonic, a prototype of AI-powered software that can turn text into audio, recognize objects, and recognize the author’s voice to quickly create sound effects and background audio for video projects.
Image source: Adobe
«We wanted to give our users control over the process, […] to go beyond the original text-to-audio workflow, and that’s why we’ve been working on an audio app like this that really gives you precise control over energy and timing and turns it into an expressive tool.” — said the head of AI at Adobe, Justin Salamon.
Super Sonic uses AI to recognize objects anywhere in the footage to create a query and generate the desired sound. In another mode, the instrument analyzes various characteristics of the voice and the sound spectrum and uses the resulting data to control the generation process. The user does not have to use their voice; they can clap their hands, play an instrument, or extract the original sound in any other available way.
It is worth noting that Adobe traditionally presents a number of experimental features at the Max conference. In the future, many of them end up in the Adobe Creative Suite. Super Sonic could be a useful addition, for example, to Adobe Premiere, but so far the future prospects of the project are unclear, and it remains in the status of a demo version.
Previously, Super Sonic developers were involved in the development of Firefly’s generative AI feature called Generative Extend, which allowed short video clips to be extended by several seconds, including an audio track.
The ability to create sound effects from a text query or voice is a useful feature, but far from groundbreaking. Many companies, such as ElevenLabs, already offer similar commercial tools.