Composing with artificial intelligence
Alongside the growth of ever-perfecting algorithms in task-specific products, we’ve also seen a groundswell in applications that can aid composition, alongside those AI platforms that can perfectly mimic human composers
Striking a balance between technical skill and inspired, creative flair is often a crucial aim when trying to launch a career as a composer. Too technical, and your work runs the risk of being perceived as soulless. But, if your music is too untethered from convention, loose, and pigeon-hole evading, then you’ll have a much harder time finding listeners.
This is doubly true when it comes to the world of professional soundtracking. Very often, those working in the soundtracking domain are provided with a brief (or pitch to one). Typically, the required music needs to fit the right ‘feel’ that the project (be it film, TV, video game or advertisement) requires. This has been the norm for professional soundtrackers for decades. Now, with the startling rise of AI-based music platforms, those patrons needing quick and cost-effective soundtracks that still sound composed, have the easy means to generate just that, for minimal outlay.
Built using detailed algorithms, platforms such as Aiva, Amper and Ecrett Music process the inputs of its users and build surprisingly stirring music at the drop of a hat. Take Ecrett Music: by selecting options from a series of menus (Scene, Mood and Genre) the smart platform efficiently concocts a piece of music that totally fits the criteria, based on a library of musicological knowledge.
Random access melodies
As we explored earlier in this feature, the road to compositional AI has had a slow and gradual development over the last few decades, but an explosion in processing power has yielded a wide choice when it comes to this new breed of digital composers. While the option of leaning on an AI to dispense ready-to-go, royalty-free tracks is certainly an attractive one for many, there are other applications of AI in the compositional landscape which invite more innovative interplay between man and machine. Google’s Magenta is among the most interesting forays into working with AI to make music and art. The open-source research project has spawned numerous plugins, such as the fascinating NSynth. With the N standing for ‘Neural Synthesiser’, NSynth’s function is to create new sounds by interpolating existing pairs of sounds which the users present to it. “Instead of using AI in the place of human creativity,” write developers Magenta, “we strive to infuse our tools with deeper understanding so that they’re more intuitive and inspiring.” These kinds of new and exploratory amalgamations of neural networks and traditional techniques are among the most exciting AI-ventures which could open up whole new avenues of exploration.
While tools such as this might be firmly assigned to the ‘experimental’ camp for now, there’s other AI-equipped tech which could provide everyday assistance in the studio. Orb Plugins’ Orb Producer Suite is a series of four composition-aiding plugins that don’t seek to take ownership of the creative process like some of the subscription services do. Instead they work to build out further developments of
your own ideas, by suggesting chord sequences, musicologically solid routes forward and seamless integration into your DAW so your own creative workflow isn’t impeded. It can also suggest new sequences to you. Almost like working with an equally musically minded bandmate. It’s a similar story with Splash PRO, a completely free and DAW-agnostic way to inject a savvy virtual compositional assistant into your life. One of its coolest features is its ability to sing back written text. It works surprisingly well.
Get smart
Beyond the musicological, other track elements – such as the rhythm section – are just as important to making a stellar end result. Working with an AI drum designer and sequencer, such as Audiomodern’s sublime Playbeat 3, can be hugely creatively stimulating.
“Our new [SMART] algorithm is designed to store personalised data independently for each user and can perform tasks that require decision-making,” Audiomodern’s Max P. Maximilianos explains to us. “In this case we’re focusing on reactive AI, which is programmed to provide a predictable output based on the data it receives from each user. It is a very different concept to the other two available randomisation modes currently available in Playbeat.” Max explains that this technology has an array of applications: “Moreover, besides porting this new technology to our other apps, our current SMART algorithm can be combined with our genre tagging system (when a user clicks to expand the preset’s additional information) and will be available with our upcoming update. In addition, we’re working on a bigger development which will allow all apps to communicate together.”
Songwriting copyright is often a minefield once it comes to the breakdown of who wrote what, and how much percentage of the track’s success should go to its principal composer and musicians. Thankfully, AI hasn’t developed an interest in getting paid for its work (yet!). So having a bit of back and forth with AI-driven creative platforms like Orb Producer Suite, Playbeat, Splash Pro or the slick, browser-based Hookpad, means you don’t need to keep notes on who wrote what, while still reaping the benefits of a genuine collaborator.