Belarusian shift worker in Yamal creates songs in Belarusian with AI. "Music used to be a closed club. Neural networks kicked open these doors"
Dzmitry Yarmolich works in the oil and gas industry: previously in Alaska, now in Yamal. But he self-actualizes in music. He considers neural networks to be instruments just like a synthesizer, piano, or wooden spoons.
Dzmitry Yarmolich works on a rotational basis in the oil and gas industry. He previously worked at oil facilities in the North Slope region in northern Alaska, then in Yamal and the Vyngapur gas field in the Yamal-Nenets Autonomous Okrug.
Dzmitry is 41 years old, lives in Minsk, is married, and has children. However, in parallel with his work on drilling rigs, he is involved in music. His current project is called "Vuž" (Grass Snake), and he releases songs in Belarusian, created with the help of neural networks. The man talks about his experiments on social media.
How the "ice melted" on the drilling rig
Dzmitry Yarmolich's musical history began long before "Vuž" and neural networks. Under the pseudonym D-i-medroL, Yarmolich released individual lyrical tracks. Back in September 2017, he published a video on YouTube called "Bur idzot" (The Rig Goes) – a parody of the hit song "Tayet lyod" (The Ice is Melting) by the band "Griby" at the time. In the video, workers in special clothing dance against the backdrop of a drilling rig, and the familiar refrain turns into a song about shift work and drilling. The video then spread widely across social networks, communities, and video platforms.
The new project "Vuž", which would appear eight years later, is completely different from everything Dzmitry had done before. Belarusian language, mythology, and history became the material for his creativity.
Initially pretended to have accidentally found a Belarusian hit
Yarmolich published the first track of the new project in Belarusian in December 2025. But he didn't immediately admit his involvement. He posted several videos as if he had accidentally discovered unknown Belarusian music and was now sharing the discovery:
"Belarus bursts into the charts of 2026! Heard that sound? That's when ancestors groove no worse than top DJs. Ethnofuturism we've earned. Proud of our people."
In other videos, he admired the combination of Belarusian music and traditional instruments with club beats:
"Belarusian music can be so mind-blowing. I'm just in shock."
The trick worked. Hundreds of mostly positive comments appeared under the videos:
"How pleasant to hear our native language."
"In my native language — I will listen to it all my life. Thank you very much."
"Goosebumps. The song is fire!"
"God, how wonderful. Ours, Belarusian."
There were also skeptical comments – including assumptions that the music was AI-generated. Nevertheless, one of the videos garnered about 200,000 views in 24 hours.
Only then did Yarmolich reveal his cards: Vuž is his own project. According to the author, he wrote the lyrics and prompts himself, and obtained the music and vocals with the help of a neural network.
"I always wanted to create something worthy in Belarusian, something you could listen to in the car, something that would pump you up. And I see that you liked it too," he explained.
"Belarusian code in digital sound"
Yarmolich started actively managing his TikTok specifically to develop "Vuž". He now has about 5400 followers. He describes his music as follows:
"Belarusian code in digital sound. The best modern folk and ethno house for your speakers."
Listeners rejoice at the mere presence of the Belarusian language in modern dance music, but occasionally notice errors in pronunciation and stress. Yarmolich calmly accepts comments, but sometimes responds to haters.
It should be noted that music generators work much better with English and Russian than with Belarusian. They can mechanically read a letter, make a mistake with stress, or make a phrase phonetically unnatural. Therefore, generated songs can often grate on the ears of those who are proficient in Belarusian.
From love to Bahnik and Rahnieda
Gradually, the project expanded beyond a few dance songs. Yarmolich began experimenting with female vocals, reworking pieces by lesser-known authors, and creating Belarusian versions of famous songs. Among these attempts was "Zialiony vinahrad" (Green Grapes) by Aliaksandr Saladukha.
A separate direction is Belarusian mythology. "Vuž" created songs and AI-generated videos about Bahnik, Lozavik, and Arzhavienik.
"I maximally intertwined the modern rhythm of life with our Belarusian mythology. I also added, of course, love — both mutual and that which tears the soul apart," the author recounted while working on his second album.
In February, he embarked on a cycle about legendary women of Belarusian history: Rahnieda, Nastassia Slutskaya, Euphrosyne of Polotsk. He named the song about Rahnieda "Krou Polatska" (Blood of Polotsk), and along with the music videos, he published brief historical references.
Another source of inspiration is his subscribers. Yarmolich invites them to send personal stories, writes Belarusian-language songs based on them, and consults about sound variations. This creates a small but quite engaged community around the project. For example, users made about 150 of their own videos for the track "Viasellie" (Wedding).
Five albums in six months
Artificial intelligence has allowed Yarmolich to release music at a pace almost impossible for a traditional band.
Thus, in six months he released five albums. This includes the album "Lios" (Fate) – 17 compositions lasting over an hour. The album "Ja — Vuž" (I Am Vuž) contains 11 songs. Later, "Bielaruskaja viasna" (Belarusian Spring), "Kropli na shkile" (Drops on Glass), and the latest August album "Nuliavy kilometr" (Zero Kilometer) with eight tracks were added.
The music is available on Spotify, Apple Music, "VK Music", "Yandex Music" and other services.
The audience of "Vuž" turned out to be roughly the same age as the author himself — and even older. In Yarmolich's Telegram poll, almost half of the subscribers were over 45 years old. Another 38% were between 35 and 44 years old. Only four were younger than 25.
"Music used to be a closed club"
The biggest dispute around "Vuž" concerns the author's right to call himself a musician at all.
Yarmolich is well aware of the main reproach:
"You are not a musician. You are just an organ grinder who turns the handle of artificial intelligence."
He responds that every new generation of creators initially faced snobbery. In his opinion, a neural network is an instrument just like a guitar, piano, or wooden spoons:
"What difference does it make where the sound is created if it hits the target precisely? How does generation differ from a beat created from foreign samples, or from a voice polished and autotuned beyond recognition? Music is what's in your soul, what makes you move."
Yarmolich calls AI an "infinite guitar with millions of strings". To get something worthwhile, according to him, one needs taste, discernment, and the ability to write precise prompts:
"A mistake in one word — and the symphony will turn into a mess. The neural network doesn't know why it's doing it. The meaning is imbued by a human. Without hands and brains, AI only churns out tons of raw garbage. The instrument doesn't make you a creator — it only scales your talent or its absence."
Yarmolich sees the democratization of talents in neural networks. Previously, he argues, hundreds of dollars were required to record a track: separately for the beat, studio, mixing, mastering, and distribution. Now, a free service or a subscription for 10-15 dollars is sufficient.
Some of Vuž's arguments appear debatable, for example, that a neural network can make sound cleaner in minutes than professionals in a studio can in several weeks. It can quickly produce a technically effective result, but it doesn't always provide natural diction, accurate stresses, original arrangement, or the ability to control individual recording elements.
The parallel between generating an entire song and using samples, autotune, or electronic instruments also seems questionable. In traditional production, a musician still makes many small decisions — creating or selecting parts, changing harmony, rhythm, timbre, and structure. A generator user also makes creative choices, but their degree of control over the final result can be significantly less. Finally, the democratization of access to music creation does not guarantee success. When songs become cheap and easy to generate, an advantage goes to those who better master social media algorithms, have money for promotion, or can continuously release content. Therefore, AI does not so much eliminate the music industry with its filters as it creates new, platform-based ones.
Nevertheless, Vuž's main idea is clear: a person without musical education and large sums of money has for the first time gained the opportunity to test their idea in an almost finished sound.
"Music used to be a closed club for those with a golden card or a good producer. No money for a studio — we don't need your talent. Neural networks kicked open these doors. Now it's not a battle of wallets, but a battle of ideas."
Why some AI songs garner millions
As evidence, Yarmolich mentions SDP, Sasha Komovich, and AltVerse – projects whose music is entirely or partially created with the help of AI.
The most successful of them is SDP, which transforms classic Russian poetry into modern songs. "Syph, Garmonika!" (Play, Accordion!) based on a poem by Sergei Yesenin, garnered over 4.3 million listens on Spotify and about 8.6 million views on YouTube. "Kakaya noch! Ya ne mogu" (What a night! I can't) has over two million listens on Spotify.
Sasha Komovich became famous thanks to the AI version of "Rasskazhi, Snegurochka" (Tell Me, Snow Maiden). The track has about 2.5 million listens on Spotify and 4.7 million on YouTube Music.
With AltVerse, the situation is more complex. On social media, this project is often attributed the viral Afro-soul version of "Papaoutai". The song indeed became a global hit and garnered over 14 million listens on Spotify in its first month.
If we consider the secret of such success, each of these projects took cultural material already familiar to the audience: Yesenin's poems, the New Year's song from the Soviet cartoon "Nu, pogodi!" (Well, Just You Wait!), or Stromae's global hit. The listener doesn't need to get acquainted with the story from scratch; they immediately recognize the words or melody but hear them in an unexpected arrangement.
The second factor is one distinct viral track. A mass audience is drawn in by a single song that lands in short videos, playlists, and algorithm recommendations.
For "Vuž", the situation is different so far, and it doesn't have such view counts.
But "Vuž" has something that many faceless AI projects lack: a recognizable theme and a living author who doesn't hide behind a fictional character. A shift worker from a drilling rig who, between trips to the Far North, writes about Rahnieda, swamp spirits, and unhappy love.
In one of the videos, Yarmolich asks the most pertinent question:
"If you take away your violin, guitar, piano, or access to AI, will you remain a musician or a creator? Or are you just a guy with a stick, reproducing someone else's work? True creativity is in the mind."
Meanwhile, listeners leave simple comments under "Vuž":
"Belarusian language — simply super."
"Vuž, you are power! To say that your tracks bring infinite pleasure – that would be an understatement."
"We put on the first track – and it was an instant hit! It was so engaging, we were driving on the highway with good acoustics, it was simply mind-blowing! Thank you, brother, for good music, and most importantly, in our native language!"
Who else creates Belarusian music with AI
"Vuž" is not the first attempt to make a neural network sing in Belarusian. Several projects already exist, but each uses the technology in its own way: one puts classical poetry to beats, another experiments with folklore, and a third turns their own poems into songs.
Most similar to "Vuž" is Setivir, an experimental project by Aliaksei Kartynnik. His catalog contains over a dozen Belarusian-language tracks, including "Rodny Floŭ" (Native Flow), "Mužyk-bielarus" (Belarusian Man), "Liatsieli" (They Flew), "Kupala", and "Bitaja darožanka" (Beaten Path). Some songs are based on folk texts and Belarusian classics. The author openly talks about using Suno and shows how he manually refines the material for several hours after generation.
The project's most popular track, "Rodny Floŭ", garnered about five thousand views on YouTube, while "Mužyk-bielarus" got about 2.6 thousand. This is significantly less than the viral metrics of "Vuž"'s videos.
Another interesting example is the Neuro-Vocal Instrumental Ensemble "Dryhviaty", created by Aliaksandr Hurnik. The author himself calls it "children's songs for adults in Belarusian." The playlist contains 18 compositions: "Dryhva" (Quagmire), "Prakliony" (Curses), "Niaveličkaje ščascie" (Small Happiness), "Sustreča" (Meeting), "Svetapohliad movy" (Worldview of Language) and others. In the prompts, Belarusian dulcimer, bagpipes, and zhaleyka meet funk, folk-metal, J-pop, and K-pop. But "Dryhviaty" currently remains more an experiment within Suno than an independent media brand.
On the border between a full-fledged project and a series of experiments is Guslee Bel — "Belarusian poems on beats". On social media, the authors publish AI-processed works and mashups, but it's not always possible to precisely determine which part of a particular work was created by AI and which by a human.
In February 2026, even the National Academy of Sciences attempted to make a neural network sing in Belarusian. During a discussion about AI's ability to work with the Belarusian language, scientists demonstrated generated songs. However, this remained merely a scientific popularization experiment.
-
A Guy Came from Yemen to Become a Dentist, and in 10 Years He Almost Became Belarusian
-
In Gomel, an old motor ship floated on the Sozh without a crew on board
-
Half were denied, a third were granted refugee status. August results of reviews of Belarusians' applications for international protection in Poland have become known
"Didn't like the products on the shelf — and they write an SMS there." The Ministry of Internal Affairs called Threads a security threat and announced arrests 5050
Comments
Пластмасавы свет перамог, макет аказаўся мацнейшым - пакуль усё ж не. Добра, што моладзі цікава на жывых інструментах пабрыньчаць.
ШІ хай будзе забаўкай, а творчасць застаецца жывой.
Очень в тему вспомнили. Хотел бы посмотреть на тот AI, который сможет повторить Летова (не спародировать, а создать что-то подобное). Вроде слушаешь - и текст хрень на грани понимания, и музыка простая... а офигеваешь, аж слезы наворачиваются. Спустя три десятка лет со времен юности - все так же, я проверял... правда, добавилась ностальгия по ней, по юности ) Уж и Егора того нет сколько годков...
Кто-то это описал как картину - ее надо смотреть с расстояния, а не разглядывать отдельные мазки краски. Думаю, до такого "понимания" музыки AI дойдет примерно никогда.