As the MUTEK team wrote in the MUSIC x newsletter last August, it is impossible to tell whether AI music is good. Not only because there are so many different applications, but also because the definition of “good music” is quite a subjective one. AI music in the broadest sense of the word is an important element in the Open Culture Tech project. We look for ways in which it can add value for artists, without harming them in their creative process.
For many artists in our direct surroundings, AI feels like a monster under their bed, ready to attack their professional career and creative abilities. Because of that, most of them are mainly interested in AI technologies that can help them with their visual performance, or the interaction with the audience. But really having AI interfere with their music is something they are very cautious with.
At Open Culture Tech, we try to look at music AI tools as Frankenstein’s monster. Whether they are voice cloning tools or generative algorithms that create melodies, chords or beats. Just as Frankenstein’s monster was built up from separate elements (limbs), AI tools can also be seen as separate elements in the musical creation process. And just like doctor Frankenstein, musicians still need creative ideas and skills to connect the separate elements and bring the whole body to life.
But when we take a look at the current market for AI music tools, there is something strange going on. Many tools that are currently available, such as AIVA or BandLab Songstarter are being promoted as tools that provide “inspiration” by generating MIDI beats or melodies from scratch. In essence, there is nothing wrong with that. However, professional artists, or artists with professional ambitions, are not the right target audience for these specific tools.
So far, we have not spoken to a single artist with a lack of inspiration or musical ideas. To go even further, many artists seem to enjoy these first steps in their creation process the most since that is the point where they are really being creative, without having to think too much about the end results yet. The idea that AI needs to help us kick-start our human creativity feels wrong. Of course, if your not a musician and you need some quick stock music for your TikTok videos, these tools could be very helpful. But for professional musicians, this is not very helpful.
Two weeks ago, Google and YouTube introduced the Lyria AI model that was accompanied by various Dream Track applications. The most prominent app allows users to enter a topic (write a prompt), choose an artist from a carousel and generate a 30 second soundtrack. The AI will generate original lyrics, backing tracks and an AI-generated voice will sing the song for you. But again, this application is aiming at social media content – mainly promoting YouTube Shorts.
When diving into other websites that showcase AI music tools, such as AI Tools Club, you'll see that the majority of applications are just not created to help professional musicians. Like Mubert, they also want to support social media, podcast or app creators.
(The other Dream Track app allows you to transform your humming or beatbox into an instrument such as a saxophone or even an entire orchestra. This is very similar to Vochlea, an application that helps musicians quickly “sketch” their ideas into MIDI).
Although AI tools that translate humming into MIDI might be helpful, we at Open Culture Tech feel that there is a growing gap between the AI (application) community and the wider professional music community. Yes, there are very interesting music AI initiatives out there such as the Sound Of AI YouTube channel (recommended!) but this is not a plug-and-play app such as AIVA.
In the end, the limbs of Frankenstein’s monster are not developed for professional artists but for TikTok creators. That is why Open Culture Tech is currently working on 3 different easy-to-use AI applications that are aiming to support or inspire professional musicians during different parts of their creative process.
Together with our collaborating artists and AI developers, we will test our first prototypes soon and share our findings with you in the upcoming updates.