This is a podcast featuring those who create content. The guest of the episode is Alexey Kochetkov, CEO , discussing generative music and his vision for the future of audio content.
- listen or
- subscribe to the podcast or
Alexey Kochetkov, CEO
: Since we talk not only about text and conversational content, naturally, we didn't overlook music. Especially a relatively new direction in this field. Alexey, you are the CEO of the project . This is a streaming service that creates generative music. Can you explain how it works?
Alexey: Generative music is created in real-time by algorithms. This is music that can be adapted, applied in any field, personalized, and so on. It is assembled in real-time from a certain number of samples.
A sample is a fragment of music that every musician has the opportunity to record. In other words, generative music is created from what are referred to in English as human-made samples. The algorithm analyzes them and creates a stream just for you.
Alina: Great. The music is created by an algorithm, and the algorithm is created by people.
It makes sense to talk a bit about the background of this project, about its beginnings. Why did you decide to get involved in it? Was it related to your musical interests?
Alexey: As they say, startups are born out of pain. I was running and constantly had a side ache from switching music. At that moment, the idea came to me: why not create an app where samples would blend into an endless composition matching my running tempo? That's how the first idea of Mubert was born.
The team was assembled that same day, and we started creating a product that later made several pivots. But the concept itself is the same as it was on the first day.
This is music that has no beginning, no end, no pauses, and no transitions between songs.
Alina: Did your musical background influence your choices or the certain steps taken during the app development process?
Alexey: No. I have a jazz music education, and it hardly helps here. I know musical notes, how to play the double bass, and what music consists of.
I have always been responsible for the bass. In all the groups I have been part of, I have always taken the lowest frequencies and played the double bass, bass guitar, and bass synthesizers. This doesnât help with my work on Mubert. I just have a rough understanding of how music works, I listen to a lot of it, and I have long believed that there is no bad music or bad taste.
There is personal taste and a personal approach to music. Each person has their own, and everyone has the right to choose music and express their taste accordingly.
A little knowledge of notes, harmonies, and the like helps me. But overall, apart from me, about fifty other musicians are working on Mubert, actively participating in the development of the interface, music ranking systems, and artificial intelligence systems. These are people who constantly give advice and influence how Mubert sounds today.
Alina: Can we say that generative music is essentially the type of music that harmoniously complements other activities?
For example, usually writing text or working while listening to music is a matter of personal preference. Some may get used to it, while others may not. Can algorithmically generated music create a synergistic effect that helps achieve a state of flow?
Alexey: This is a hypothesis, and we are trying to test it.
Generative music will soon be featured in an app we are developing in collaboration with Bookmate. People run marathons while listening, and it's the only app that allows you to run without changing your pace for four, eight, or sixteen hours, and so on. People also work and study with this music. This might be a cool approach to musicâsponsoring your passion. But this is just a hypothesis.
Alina: And you are testing it through collaborations?
Alexey: It is confirmed by subscriptions and streams that happen every day on Mubert. For instance, meditation is our most purchased channel.
There are three paid channels: Meditate, Sleep, and High. High is for dub and reggae. The most purchased is meditation because during meditation, the music should not stop or change. Mubert performs this.
Alina: And what is High for if we don't take it literally? (laughs)
Alexey: To relax, unwind, feel some kind of connection, and so on.
Alina: Great. Tell me, do you think generative music â algorithmic, repetitive, long-lasting â is something fundamentally new or a continuation of ethnic, shamanic, and meditative music?
Alexey: It's something like a repeat.
Mubert actually started back in 2000 when I was recording from Monte Carlo radio [track] . As soon as it played on the radio, I kept recording it on a cassette until I filled an entire side with that track. Then I did the same with the other side. In the end, I had a whole cassette that had only Bomfunk MCâs â Freestyler recorded on it.
Mubert brings back those times. A lot of people use music on repeat. They play a certain track and work to it all day or engage in sports for a while.
Generative music in its current state doesnât quite have the drama that a DJ can provide. The DJ understands in real time what needs to be ramped up , what needs to be lowered, expanded in harmony, or narrowed. Generative music only strives for that.
And we are pioneers in creating drama in generative music, which we have learned to make endlessly long, smooth, and coherent. We are currently learning to create drama within it.
As we recently demonstrated at the adidas store. We created a DJ set without a DJ, and a lot of people danced wonderfully to this music. It sounded at the level of German DJs who were the original sample creators. But the set itself was created by Mubert.
In response to the question, generative music traces its roots back to repetition and will end up with things we cannot yet imagine.
Alina: How does the algorithm work?
Alexey: The algorithm analyzes many parameters: melodiousness, rhythmicity, richness, sound 'fatness', instrument. Its tempo, key, and so on. A lot of parameters that are objective. Then we start with subjective parameters. This includes genre, activity, your taste. There can be parameters related to geodata. For instance, when you want to gather an urban stream â understand how the city of Berlin sounds.
The AI system here is an assistant to ensure that subjective parameters are fulfilled. So that you can receive music during your activities based on your taste and on the things you have already shown in this system.
Soon we will release an app where you can like, dislike, and favorite music to influence your personal style. This will be the first app in the world without a general chart. We don't even have a concept of a general popularity or unpopularity chart for samples and artists in our database. Each user has their own chart containing combinations of parameters. Based on these, the system learns and creates your personal soundtrack.
Alina: Essentially, we are saying that for every user, Mubert generates multiple soundtracks for different aspects of their life.
Alexey: Yes. This is the first true personal streaming.
Alina: Great. You mentioned the collaboration with Adidas; can you please tell us about collaborations with brands in general? How do they look?
Alexey: Music is the closest form of creativity to humans. Therefore, if a brand wants to connect with people, it should do so through music. Few are aware of this yet, but those brands that do are starting to act on it.
For example, Adidas holds pop-up parties that occur unexpectedly in their stores. They are not advertised. Other brands sponsor themed parties.
Who else should they turn to but new technologies? They have two options: either hire a top DJ or use top technology. If there's a way to combine both â as we did with Adidas when a top Berlin producer provided the samples. â a person who created electronic music. Then the brightest spark is born, shining so that the brand can make a statement about itself.
For any brand, music is a reason for information dissemination.
Alina: If we talk about parties... Naturally, there are a lot of people. How does Mubert understand what music to play? How does personalization work in this case?
Alexey: The party is customized for the event, the city is customized for the city. That's allâŠ
Alina: The essence.
Alexey: Yes, itâs an entity that we can customize. Personalization ranges from your time of day and date to various global aspects. As I explained earlier, there are objective parameters and there are subjective ones. The combination of subjective parameters includes genre, city, you, morningâanything. The objective parameters consist of sound density, tempo, tone, gamma, and so on. All those elements that can be measured objectively.
Alina: How do you think generative music and music overall will develop? Will algorithms eventually replace human composers or DJs?
Alexey: By no means. The DJ-selector will remain. You can't assemble music better than a DJ canâwhether it's track-based or sample-based music. DJs were once called selectors, and that role will persist because they bring together the best parts.
The development of generative music will lead to it being on every phone, as it offers different opportunities for adapting and personalizing music. It will also feature curated selections. For instance, we could exchange some generations and understand how you trained your Mubert and how I trained mine. It's somewhat like today with playlists, but on a deeper level.
Alina: So, the future of generative music is a symbiosis between the human creator and an algorithm that analyzes everything happening more deeply and accurately?
Alexey: Absolutely.
Alina: Great. Lastly, our quickfire round of two questions. Music helpsâŠ
Alexey: Live, breathe.
Alina: The best track is oneâŠ
Alexey: That really hits.
Alina: Awesome, thank you very much.
Our microformat on content marketing:
P.S. In the profile are links to all episodes of our podcast.
Source: habr.com
