Magazine \ Features \ Features

AI and music, part 2

How AI is changing the way music is produced

2019 Aug 24     
2 Bit Thugs

From composition to mastering, there are now AIs to help with every step of the music production process, but is that a good thing?

In our last article about Artificial Intelligence (AI) and music technology, we looked at its possible impact on consumers. Now we turn to the music makers, producers and mastering engineers for their take on AI. 

Renegade masters
Perhaps the most noticeable growth area of AI in music technology has been mastering services like LANDR and Izotope, which use AI rather than a human engineer. 

LANDR have always said their product is for those who couldn’t otherwise afford a mastering engineer and isn’t intended to replace them – rather, it’s an attempt to create a new market. However, critics of AI mastering argue that it’s just another digital shortcut, like an Instagram filter or the CDJ sync control. AI mastering is sometimes seen as part of a larger modern malaise of laziness and instant gratification, where accessibility leads to the devaluing of skills and the abandoning of expertise. 

Rob Small has mastered music by Ricardo Villalobos, Paul Woolford, Hot Since 82 and many more, and feels that AI mastering "lacks the human emotion that comes with listening to music, which is a crucial part in mastering, production or any form of artistic creation".

"AI mastering helps show what a great engineer can do with your music, rather than just make it louder and brighter," he continues. "The dialogue that comes with the session, the advice an engineer can give artists before mastering takes place, the feedback on the music post-mastering – all of these things are just priceless and AI mastering can't and won't provide them, because is it NOT listening to your music. It's not able to think about how the artist intended the material to sound, or whether or not something is intentional sound design."

However, many producers, such as Danny Lewis (Ministry of Sound, Ruff Trx, Plastik People, pictured above), have a different take. "I'm very low on time for getting stuff out and need to focus on composition and mixdown," says Lewis. "Often I can't play stuff out loud when I need to and need a way of validating, and improving where necessary, the finished products. I really like the results compared to my DIY masters, and having the flexibility to adjust the settings should they need a little tweak. Sometimes it will give me a new perspective on something I've done and allow me to go back to the mixdown and then master again. I think it's a great application of AI."

Compositional assistance
Aside from mastering, AI is being used for compositional assistance, too. Orb Composer is an AI that contains a vast database of musical information based on millions of pieces of music. The user can try out musical ideas or Orb can generate its own chord progressions and melodies in different styles and moods based on guidelines you give it. And it can be used without any music theory knowledge. But aside from basing compositional decisions on parameters you enter, AIs are now being developed that can also observe your compositional and production preferences over time and make suggestions based on them.

Andrew Emil (pictured below) is a US producer, musician and DJ who uses various AI tools. He told us: "I only see intelligence-based processing continuing to assist composers, producers, singers and instrumentalists with more and more utilities that will begin to learn your preferences, and then begin to anticipate decisions in your creative process, offering up compositional suggestions, harmonic motion choices and any other pattern-based suggestive metric compiled from your entire body of work, based around data sets it has access to, like DAW sessions, manuscripts, and other personal settings and workflow measurements."

Danny Lewis argues: "I think AI in music is only going to improve. I've worked with data scientists over the years who do incredible things once they have data at scale. If we are able to data-mine the entire compositional landscape, I'm convinced we'll get accurate mastering 'soundalikes' in the future. DAWs will offer that as a feature – within the next 10 years, you'll probably be able to pay a small fee to master your tunes to sound like Kerri Chandler!"

Currently, this positive take on AI tends to be limited to those working with it – there’s plenty of opposition from those concerned that AI could replace musicians and producers. Distrust of new technology has, of course, been a constant theme in the music industry: musicians worried that the wireless would render them unneeded, that samplers would make them redundant or that home taping was killing music. In each of these cases the concerns were valid – new technology can cause substantial change, often impacting certain job roles – but the net effect is always complex, bringing new opportunities and benefits too. 

Even Rob Small sees a positive side: "AI puts great mastering engineers in a really good position because it shows what we can do and makes sure that we're at the top of our game… I actually receive more work these days thanks to AI mastering, because it's showing what a proper engineer can do with your creations and how we can improve your final creative output." 

Early adopters believe that AI technology will empower creators and give more people access to music creation. Mau5trap producer Dom Kane says, "I think the more you understand something like this, the more you realise that taking over human jobs and wiping them out just isn't as easy as it sounds. I definitely think AI is more exciting than scary."

Andrew Emil again: "Remember when ALL DRUMMERS ARE GOING OUT OF BUSINESS when drum machines came out?! I, like most musicians, composers and producers, welcome tools that will help us achieve the ultimate end game results as quickly as possible. I am here to make good music, as often as possible. I don't care if it was made on a vintage synth or a new plug-in, it’s only what’s coming out of the speakers that matters."

Back to the future
Looking forward, AI has the potential to make huge changes to music creation and composition. Sound artist Holly Herndon recently co-developed a machine learning AI to recreate the sound of a particular human voice, where the AI is treated as just another member of her musical group, a machine collaborator. The results, an intriguing collection of melodic syllables and sounds that have no literal meaning but are perfectly musically valid, can be heard on her PROTO album, released earlier this year. 

The idea that AI will simply replace people oversimplifies a much more complex process. Instead, roles will change and evolve: some may fade, others come in their place and new opportunities are created all the time, even as older traditions are left behind. And anyway, as Andrew Emil observes:  "Automation is going to do away with jobs that can be automated, but it’s not going to take away creative ventures. What would even be the point in that? You automate processes that are redundant, you don’t automate activities that bring you pleasure."

There is no doubt that machines are getting good at certain parts of the production and composition process. But, at least for the moment, there are certain essential human contributions that simply can’t be boiled down to a set of algorithms or machine learning. It looks like the robots aren’t going to take over the world just yet – but one of them might write a number one hit soon. 

Words: Harold Heath Main pic: Gerd Altmann/Pixabay





Tags: AI, artificial intellegence, LANDR, Izotope, Danny Lewis, Dom Kane, Rob Small, Andrew Emil, Holly Herndon, Orb Composer