AIVA: Composing In A(I) Minor

AIVA co-founders Denis Shtefan and Pierre Barreau (Photo: AIVA)

While ChatGPT has simplified many tasks, it has not figured out how to compose music (yet). AIVA, the Luxembourg AI-powered music composition platform on the other hand has been doing this with increasing levels of success since 2016.

Launched in 2016, the startup spent its first years deeply involved in R&D before offering its solution as a service. While the solution initially only focused on classical soundtrack music, it has since expanded to include jazz, electronic and hip-hop music, among others.

First tailored towards those active in the industry, the platform underwent some upgrades in June last year, enabling users with little to no tech knowledge to create their own musical compositions.

“Instead of just having the user select some very broad style category that we trained our system on, now the user has full control to define those stylistic categories and really be part of the full process to make it more creatively enjoyable,” said CEO Pierre Barreau. 

Goodbye, writer’s block?

AIVA, which stands for Artificial Intelligence Virtual Artists, was fed over 30,000 musical scores “of history’s greatest”. Using neural networks, AIVA learned to find the patterns in these scores which allows it to infer what comes next from a couple of bars of music alone.

Instead of wanting to replace the person in need of the tool – something every AI developer says – AIVA aims to support composers and amateurs in their work. It is especially useful in the early stages when composers are staring at a daunting blank page and are struggling to make progress.

“A composer can use AIVA to generate a few compositions and ideas quickly and then shape them according to his taste and needs instead of starting composing from a blank page,” said Ashkhen Zakharyan, sales manager at AIVA. 

Interestingly, composers can choose from a predefined list of styles or upload their own to the platform and see what AIVA makes of it. The flexibility the solution offers is impressive and allows the user not only to determine the song style but also the emotional impact, length and type of composition.

Personalised music

While many professional composers have already used and created compositions with AIVA – Dana Leong (a Grammy-winning composer) and pop singer Taryn Southern are two examples, the solution is still mostly geared towards amateur creators and composers on a budget.

However, AIVA is now looking to expand its reach towards personalised soundtracks based on individuals’ music preferences. 

“That’s been an idea that’s been resonating with us and is part of our company mission,” said Pierre.

The 15-person-strong company, which last raised €1.5m in June 2020 with China’s second-biggest music streaming provider NetEase, is currently mostly active on the Chinese market. As one of the most mature and well-funded AI-driven composition platforms, the CEO feels that they are “in a pretty good situation right now.”

This article was first published in the Silicon Luxembourg magazine. Get your copy.

Related Posts