A Star is Born – A KLU start-up takes off
Lacking the technology and resources enjoyed by heavyweights such as Warner and Universal, independent music labels are often forced to manually search online platforms for new talent. It is a time-consuming, often random process; yet AIDAR (AI driven A&R), an innovative start-up founded by KLU alumnus Dr. Janek Meyn and PhD student Yana Asenova, uses artificial intelligence to help smaller labels discover the stars of the future.
The idea for AIDAR, co-founder and KLU doctoral graduate Dr. Janek Meyn says, has its origins in a chance meeting three years ago.
“I was doing my PhD in music marketing, and we presented our work on streaming service remuneration models at Hamburg City Hall. The CEO of a local indie label was interested and said we should talk. We said, of course, because we needed consumer data for our research.”
That conversation evolved into a relationship and eventually to thoughts about how music industry data could be employed in the constant quest for the next big thing.
“Even today, many A&R – artist and repertoire – managers search for new talent by opening Spotify and randomly looking for artists,” says Meyn, whose career features stints at Lufthansa Technik and consulting giant CGI. “When I saw that, I thought let's search for artists in a more data-driven way, using machine learning to get really good fits for each label.”
Given roughly 60% of all music labels have less than five employees and annual revenues under 250,000 euros, it’s not surprising that small labels often resort to combing through online platforms for artists to sign.
“They don’t possess resources like the big three - Warner, Universal, and Sony - who have heavily invested in data and scientists,” says Meyn, who hails from Hamburg. “We want to be the data scientists for the indie labels.”
Such a project, of course, needs plenty of support to get off the ground, and KLU has furnished the start-up with not only an office, but invaluable connections, including between Meyn and co-founder Yana Asenova.
The 29-year-old Bulgarian, a data scientist and graduate of the International Music Business School in Barcelona, met Meyn only weeks after starting her PhD in music marketing at KLU in January but believes the culture of community fostered by university was instrumental in the pair teaming up.
“The institution made the distance between each other very short, meaning that even if I didn't know Janek before, some kind of trust already existed based on the fact that we both belong to KLU.”
Financial funding, meanwhile, looks to have been secured with the help of the University of Hamburg and Professor Michel Clement, who Meyn studied under during his master’s and worked with on research papers while doing his PhD. The money, granted by German funding organization EXIST, will cover salaries for a year and enable the start-up to add two other KLU graduates - Lyubomir Kushev and Caspar Hoeyng – to its staff later in the year.
Add to this an ever-growing database of artists and the successful launch of a minimum viable product (MVP), and the start-up is making great strides in the right direction.
Naturally, there’s been some resistance to the use of artificial intelligence from within the music industry, particularly from A&R reps, who worry about being rendered obsolete.
“We definitely don't want to replace them, AIDAR is just a tool to help reps do their jobs,” assures Meyn. “Right now, we have a database of 3.2 million artists with information about what kind of music they make, what stage they are in their career, how many people follow them and so on.
“We also have information about the labels, such as if they're looking for specific genre or a wider spread. Do they just want tracks which have a high danceability, for example, or which are really positive?”
Machine learning is then employed to generate the best matches of artists with labels.
“Once the label receives our recommendation, they do their normal A&R work and see if they like the artist,” explains Asenova. “If so, they get in touch with them, if not they assign the artist a lower rank on a scale of one to seven. This feedback is then used to train the machine learning model. The more it’s used, the better the AI is able to make precise recommendations to labels.”
In the way technology, from the electric guitar and drum machines to laptops and reverb software, has enriched the way music is made, it seems only logical that an innovation such as AIDAR will enhance the way those who make it are discovered.