Musiio is revolutionizing music categorization with AI
ANOUK DYUSSEMBAYEVA | AUGUST, 12 / 2021
Photos provided by Xiao'an Li
Musiio is a Singapore-headquartered music technology company that specializes in AI-driven solutions for the music industry, which includes metadata tagging at scale, audio reference search, and a playlisting product. In addition to these, it also has a Hit Potential Algorithm, which scores music based on its likelihood of commercial success — this is currently being used primarily in custom projects with select clients.
The initial idea was developed by Hazel Savage and Aron Pettersson during Entrepreneur First, an incubator located in Singapore. Aron was interested in solving the most difficult problem he could find, and Hazel — having many years of experience in the music industry for companies like Shazam and Pandora — wanted to make music categorization digital. If before there were 10-20 singles put out and people could categorize them manually, it became an impossible task as the number of songs released rose to tens of thousands. "They set out to create a solution where we would be able to use AI to provide music metadata tagging services on a large scale," begins Xiao'an Li, Musiio's music strategist. At the moment, the algorithm is able to go through five million songs per day and provide information like genre, mood and tempo, to name a few.
Because Hazel has been in the music industry for a long time, the first clients came from her ability to penetrate that market, as Xiao'an says. In the beginning, they mostly targeted sync libraries and record labels, but the music strategist now notices a slow shift toward commercial music. He also mentioned that Musiio has worked with folks in the advertising and gaming industries. While the initial code was created on Aron's laptop, the company now serves its customers using Google's Kubernetes. Since the early days, the team grew from six people to 30 in the office.
The initial idea was developed by Hazel Savage and Aron Pettersson during Entrepreneur First, an incubator located in Singapore. Aron was interested in solving the most difficult problem he could find, and Hazel — having many years of experience in the music industry for companies like Shazam and Pandora — wanted to make music categorization digital. If before there were 10-20 singles put out and people could categorize them manually, it became an impossible task as the number of songs released rose to tens of thousands. "They set out to create a solution where we would be able to use AI to provide music metadata tagging services on a large scale," begins Xiao'an Li, Musiio's music strategist. At the moment, the algorithm is able to go through five million songs per day and provide information like genre, mood and tempo, to name a few.
Because Hazel has been in the music industry for a long time, the first clients came from her ability to penetrate that market, as Xiao'an says. In the beginning, they mostly targeted sync libraries and record labels, but the music strategist now notices a slow shift toward commercial music. He also mentioned that Musiio has worked with folks in the advertising and gaming industries. While the initial code was created on Aron's laptop, the company now serves its customers using Google's Kubernetes. Since the early days, the team grew from six people to 30 in the office.
"Imagine if you've been manually doing this [categorizing music] for years, and now you have this [Musiio software]. AI is meant to give you more time for the human stuff — nobody needs to go through tens of thousands of songs and tell you what tempo each song is. Why not let a robot do that?"
The company comprises three teams that specialize in sales, music and tech, respectively. Li works with people from all three. He's also the first person at the company with a degree in music. While the community is very diverse, with employees coming from Russia, Singapore, Sweden, England, America, India and France, and is very different both in skillset and background, Musiio isn't planning to hire anyone that is strongly qualified in music. "I think they would be more interested in an all-rounder who's also able to communicate with the tech and sales team," Xiao'an shares.
At the same time, that doesn't mean that you can't combine music and tech in your education. One of Musiio's past interns, Anav, is currently studying both composition and computer science. "For people who are graduating in music from college, it's really important to develop skills in other areas because very rarely will you find the role 'music subject meta expert' in a startup, even a music startup," the composer admits. "If you have some skills in data analysis, business development or marketing, you might find yourself a little more in demand." Xiao'an himself ran two businesses — he used to run an orchestra he created — so when he was appointed for this role, it became evident that he was capable of analyzing data and building data analysis systems.
At the same time, that doesn't mean that you can't combine music and tech in your education. One of Musiio's past interns, Anav, is currently studying both composition and computer science. "For people who are graduating in music from college, it's really important to develop skills in other areas because very rarely will you find the role 'music subject meta expert' in a startup, even a music startup," the composer admits. "If you have some skills in data analysis, business development or marketing, you might find yourself a little more in demand." Xiao'an himself ran two businesses — he used to run an orchestra he created — so when he was appointed for this role, it became evident that he was capable of analyzing data and building data analysis systems.
Over the years, data analysis has been a growing field that intersects a lot of fields, not only music. "Although you don't necessarily have to have a data science degree or use SQL, it's important that you're able to look at the data and ask interesting questions about it," Li says, adding that this is how he started. If your school offers electives in data analytics, he suggests you do it, because even the more traditional music companies are becoming increasingly data driven.
"The more useful skills that you can be known for, the more likely you're going to snag the best opportunity."
When talking about what is it that music institutions lack in the way they prepare students, Xiao'an mentions basic business concepts and learning how to be a human being outside of the academic environment. "Many of the professors, with all respect to them, don't necessarily all have great experience in making a living in the real world, especially if they have been in academia for a long time," he elaborates. "I had a professor that told me: 'why don't you go to LA and knock on some doors?' I was like: 'it's not going to happen.' You get a lot of antiquated advice." Whenever you feel like something is extremely academic in college, the composer advises you find the time to meet people outside of the music field, as well as learn things that are not specifically related to your major in order to be prepared for the real world.
One of the current issues is that institutions don't give young artists a chance to be entrepreneurial. When problems are thrown at people, they will develop necessary skills in the process of solving those challenges. A little bit of math: let's say you came up with a new business idea every week for four years, at the end of that, you would have about 208 business ideas. By the time you graduate, you would most likely be good at generating ideas, especially if you were supervised and someone gave a reasonable explanation as to which of those ideas were good, bad and gave you suggestions for what to try next time.
One of the current issues is that institutions don't give young artists a chance to be entrepreneurial. When problems are thrown at people, they will develop necessary skills in the process of solving those challenges. A little bit of math: let's say you came up with a new business idea every week for four years, at the end of that, you would have about 208 business ideas. By the time you graduate, you would most likely be good at generating ideas, especially if you were supervised and someone gave a reasonable explanation as to which of those ideas were good, bad and gave you suggestions for what to try next time.
"Schools just have to create opportunities for students to start devoting brain energy in the direction of entrepreneurship."
Another important factor of success is your network. The only choice is to network "like hell," as Xiao'an puts it. Hazel's connections, for instance, are still one of the key ways Musiio finds its customers, along with other methods like the media visibility Musiio has had in the past years, and events. "[It's either] using existing networks or go[ing] into new areas by getting introductions or entering conferences," he tells me. "The only way to get there is to develop a reputation through writing, content and successful case studies."
Content is a big part of what Xiao'an does not only at Musiio but also career-wise. When it comes to the company's blog, however, he focuses on articles that have some kind of business implication for the target audience. Unbundling topics like how record labels can use data to better increase the likelihood of success in particular markets by using the startup's AI to study the tags that come from the popular sounds in those markets, the music strategist provides interesting insights that people can translate into business decisions.
Recently, Musiio released a case study for Vans, which holds a competition called Musicians Wanted. "This year, they got 10,000 entries; our job is to help them set up the system for determining which of these are the best," Li says. When asked whether this project is the most exciting, he laughs. There aren't a lot of things that excite him. "What I'm excited about is when somebody comes to see the technology, they talk about the problems that they have and then we find [how] to help them in a way they wouldn't have been able to achieve themselves… that's pretty exciting for me," Xiao'an concludes.
Content is a big part of what Xiao'an does not only at Musiio but also career-wise. When it comes to the company's blog, however, he focuses on articles that have some kind of business implication for the target audience. Unbundling topics like how record labels can use data to better increase the likelihood of success in particular markets by using the startup's AI to study the tags that come from the popular sounds in those markets, the music strategist provides interesting insights that people can translate into business decisions.
Recently, Musiio released a case study for Vans, which holds a competition called Musicians Wanted. "This year, they got 10,000 entries; our job is to help them set up the system for determining which of these are the best," Li says. When asked whether this project is the most exciting, he laughs. There aren't a lot of things that excite him. "What I'm excited about is when somebody comes to see the technology, they talk about the problems that they have and then we find [how] to help them in a way they wouldn't have been able to achieve themselves… that's pretty exciting for me," Xiao'an concludes.
