MUSIC TECH STARTUPS: What if a computer created the next hit song?

Startups in the music industry are nowadays flourishing, trying to answer some of the questions that the same technology that allows their existence is posing. Data security, secondary ticketing, royalties, hit music: every topic has someone covering it in more or less successful ways.

The social media help ask these questions but also spread the word about these startups, while many of them get a kickstart from incubators. Marketplus will run a special in the following weeks regarding these startups. 

This week Marketplus will look into Popgun, a company creating artificially intelligent software that will be able to play and create music.

Jack Nolan, BD at Popgun, explains how important taking part in accelerator Techstars was for them: «It has been phenomenally valuable for us. As an early stage company, it provides us access to mentors, industry connections, and connections to other startups in the space. Navigating life as an early stage start-up can be difficult, and accelerators and incubators make it much easier».

Techstars also provided monetary help. Is this or the non-monetary one more valuable?

«The non-monetary support is far more valuable than the cash. Techstars, in particular, provides access to world-class educational resources and to mentors who have been through it all».  

Do you think that in the music field there’s even more need for this?

«The music industry has historically been resistant to innovation and technological change. Having an accelerator allows startups to skip the red tape and bureaucracy and get in meetings with the people that matter. I definitely think there's a need in this space and many others».

Is Australia – and in particular Brisbane – a lively environment in this sense?

«Brisbane is growing into a hub for startups, and more recently for Machine Learning specifically. One of our founders, Adam Hibble, also founded Code Network, a community of software engineers based in Brisbane that now numbers in the thousands. Creating an environment that encourages innovation and collaboration is helping to propel Brisbane to the forefront of AI research. We're very fortunate to have hired a team of extremely talented engineers exclusively from Brisbane so far».

Talking more directly about Popgun, could you be more specific about the idea you’re carrying on?

«Popgun is creating deep learning software for the music industry. A neural network simulates the way neurons in the brain work. It can get very complex but an easy way to think of it is that it tries to learn the same way humans do. As an example, take self driving cars. You can't teach a car to drive by giving it the road rules and expecting it to know what to do. It has to learn by watching thousands and thousands of hours of driving footage and then going out and practicing. We think music is the same. You can't teach people music with a rule book, you teach by letting them practice. We think teaching a computer to play music is fundamentally the same».

What will be the impact of applying technology/AI to music?

«We believe AI will have a serious impact on every industry. Think of how much the world changed when the internet became widely available. We think AI will have an equally significant impact on every aspect of our lives. This has already begun with IoT technology leveraging Natural Language Processing and with the advent of self-driving cars».

How can you understand with algorithms how much of a hit a song is? 

«We don't think there is an algorithm for top hits, but we think they might share a few key traits. Our software will be able to better identify what these traits are, and just maybe predict which trait will define tomorrow's big hit».

What do you think will be different in your case than the other songs written by AI, and corrected by humans? How long of a process will this be?

«Our music will be different because it's not completely backward looking. We aren't teaching our software the rules of music. We're starting with the basics of what music sounds like and working from there. We don't know yet exactly what it will create but we think it will be something special». 

One thing that is often said is that creativity is peculiar of human beings. Do you think that your research demonstrates something different? Is AI creating a song that could be judged “creative”? 

«We think creativity is difficult to define. Will our software ever completely replace the human element in creation? Probably not anytime soon. But will it do things which have historically been regarded as 'creative' in nature. Definitely».