The use of artificial intelligence to create music it is not something new, but the advent of the new AI models it changed everything, and Adobe wants to stand out in a market where several tools are about to arrive that can help creators create original melodies (do you know what the best apps for using artificial intelligence are?).
In fact, a few days ago it was held in Brooklyn the Hot Pod Summitduring which the company presented Project Music GenAI Controla platform capable of generating audio starting with text descriptions or a reference melody, just like Firefly or Midjourney can create Images.
But the peculiarity of this tool, unlike similar products like MusicLM of Google or AudioCraft of Meta, which stop here, leaving the modification part to other possible programs (or starting from scratch), Adobe proposes a solution that presents itself as a sort of Photoshop for music, allowing modification in the same workflow.
Project Music GenAI Control it is currently in an early stage of development, so it is not yet available to users, but Adobe has published a video where we can see how it works.
Basically you enter a text prompt to create a melody, like “soundtrack for a film“, or “sad jazz music“, or “upbeat dance music“, and after getting a result you can adjust aspects such as time, intensity, the models repeated and the structure.
Or you can take a track and extend it to a length of your choice, remixing it or creating a loops infinite (useful for example for those who need a melody to use as background), all without the need to cut to create intros or final parts.
Adobe also states that the tool can adjust the generated audio”based on a reference melody“, which further expands its potential.
Unfortunately it isn’t there yet a user interfaceand from the video you can see the operator acting only from the command line, but we imagine that once ready, Project Music GenAI Control it will be very simple to use.
The instrument is being developed in collaboration with the University of California and Carnegie Mellon University’s School of Computer Science, but there is one interesting aspect, which is the data on which the model was trained.
This type of products are often quite opaque from the point of view of copyright, and only recently are developer companies showing greater sensitivity on the subject (also to avoid lawsuits). The concern is all the stronger for the music, where the labels they are quite powerful and were quick to take legal action.
Adobe claims to use public domain content for its AI models, and this also applies to the public demo of Project Music GenAI Controlso we expect it to also apply to the final version of the tool, but we have not had confirmation.
Adobe’s head of audio and video AI research, Gautham Mysore, said during the Hot Pod Summitwhich Adobe has a lot of attitude responsible about it, and that it is working on technology to help identify the audio produced by Project Music GenAI Control.
At the moment the legislation on copyright for contents generated with AI it is not very clear, but certainly helping to recognize them from those created by humans is a topic of great interest. Not just for the music.
As we anticipated, Project Music GenAI Control it is still in early development and we don’t know when it will launch. If you are interested, you can see its potential in the video below, while on Adobe Labs you will find all the AI tools that Adobe is preparing (Comic Blast is truly incredible).
