Mubert: how musicians and AI can work together
ANOUK DYUSSEMBAYEVA A MAY, 21 / 2020
Alan Turing generated music on a machine back in 1951, yet artificial intelligence truly began writing its own compositions only 60 years later. In 2016, Sony introduced an AI-generated track "Daddy's Car", created by the AI software Flow Machines, and the IBM Watson supercomputer composed an emo song "Not Easy" in collaboration with Alex Da Kid.
Photos provided by Alexey Kochetkov
The same year, Mubert — a music-tech startup that generates music with the power of AI — was founded. Whether you are an ardent YouTube vlogger, run a fitness app, or own a restaurant, the app gives you the opportunity to use their "millions of minutes of copyright-protected music" that it creates everyday.
"The idea behind Mubert was born on a run. As a passionate athlete, I used to make 10 to 20 km every day. The only things that could annoy were making up playlists and switching tracks. It breaks up the pace," Alexey Kochetkov, the CEO and founder of Mubert, begins. Trying to solve the problem, he decided to combine millions of music samples into a single never-ending stream of one's preferred genre and pace.
Having a background both in computer science and jazz, and owning a creative agency that worked with brands like Nike and Adidas at the time, it only seemed natural to bring these passions together.
The industry is bustling with startups that embrace the "fail fast" concept, a philosophy that is built on extensive testing of whether an idea is valuable and needed. "We have started Mubert [on] the same day the idea came to me," Alexey notes. "There were five of us: a developer, a musician, a marketer, the devops and me." Coming up with a plan, the team had a working prototype ready in half a year.
"The idea behind Mubert was born on a run. As a passionate athlete, I used to make 10 to 20 km every day. The only things that could annoy were making up playlists and switching tracks. It breaks up the pace," Alexey Kochetkov, the CEO and founder of Mubert, begins. Trying to solve the problem, he decided to combine millions of music samples into a single never-ending stream of one's preferred genre and pace.
Having a background both in computer science and jazz, and owning a creative agency that worked with brands like Nike and Adidas at the time, it only seemed natural to bring these passions together.
The industry is bustling with startups that embrace the "fail fast" concept, a philosophy that is built on extensive testing of whether an idea is valuable and needed. "We have started Mubert [on] the same day the idea came to me," Alexey notes. "There were five of us: a developer, a musician, a marketer, the devops and me." Coming up with a plan, the team had a working prototype ready in half a year.
Photo of Alexey Kochetkov
Although Mubert started out as music for running, it now provides licensing services for both businesses and individuals, which makes it very versatile. "Apps use Mubert to make the user experience better: help people run faster, sleep deeper, and work with maximum productivity. Venues use Mubert to soundtrack their spaces, [and] advertisers and video editors use Mubert to soundtrack their content," Kochetkov mentions. According to Crunchbase, the startup also reduces B2B licensing costs and eliminates legal risks because of its global top 5 proprietary database of copyright-protected samples.
From 2018, Mubert's monetization model is based on its B2C app. Three months ago, however, the main revenue stream shifted once the team started to integrate their private API to third-party apps and services. "We grow thanks to our third-party integrations, where Mubert is used millions of times per month," Alexey elaborates.
While there are a lot of questions about whether AI will replace musicians, the founder thinks this shouldn't be a concern. Instead, people should be thinking of ways to enhance the user experience. "We should work together. The number of tracks is not important, we can generate millions of tracks. The user experience is. Each particular listener is important. With the power of AI we can create a 100% personalized experience for everyone," he writes. The idea of collaboration is embedded into the startup, which allows musicians to upload and monetize their sample packs that will be used for music generation. In fact, the team is working with hundreds of composers on a regular basis.
Last year, Björk teamed up with Microsoft and a hotel in New York to make AI-generated music that changes based on current weather. To be precise, 17 years of her work were uploaded to Microsoft's AI, which then "listens" to the sensors on top of the hotel to adjust the soundtrack, depending on the weather.
Almost all of the team members are involved in music, and passion for music is, as Alexey states, "a must-have". When asked whether students majoring in music should be having mandatory classes about AI and data science, he answers with a definite yes. The company is looking to hire more music school graduates, and is closely working with a number of music schools, offering internships and lectures. "I hope that someday we will be able to create our own AI music/tech foundation to support students globally and help them to find the job in our industry," the CEO shares.
From 2018, Mubert's monetization model is based on its B2C app. Three months ago, however, the main revenue stream shifted once the team started to integrate their private API to third-party apps and services. "We grow thanks to our third-party integrations, where Mubert is used millions of times per month," Alexey elaborates.
While there are a lot of questions about whether AI will replace musicians, the founder thinks this shouldn't be a concern. Instead, people should be thinking of ways to enhance the user experience. "We should work together. The number of tracks is not important, we can generate millions of tracks. The user experience is. Each particular listener is important. With the power of AI we can create a 100% personalized experience for everyone," he writes. The idea of collaboration is embedded into the startup, which allows musicians to upload and monetize their sample packs that will be used for music generation. In fact, the team is working with hundreds of composers on a regular basis.
Last year, Björk teamed up with Microsoft and a hotel in New York to make AI-generated music that changes based on current weather. To be precise, 17 years of her work were uploaded to Microsoft's AI, which then "listens" to the sensors on top of the hotel to adjust the soundtrack, depending on the weather.
Almost all of the team members are involved in music, and passion for music is, as Alexey states, "a must-have". When asked whether students majoring in music should be having mandatory classes about AI and data science, he answers with a definite yes. The company is looking to hire more music school graduates, and is closely working with a number of music schools, offering internships and lectures. "I hope that someday we will be able to create our own AI music/tech foundation to support students globally and help them to find the job in our industry," the CEO shares.
