«Why don’t you cut the cute act, Danny, and tell me exactly what it is you’re looking for?»
«Flaws.»
«There hasn’t been a murder in six years. There’s nothing wrong with the system. It’s perfect.»
«I agree. The system is perfect. If there’s a flaw, it’s human. It always is.»
Anderton and Witwer in «Minority Report»
I stumbled upon a music video by VYRMA and something struck me as strange. The music is catchy, but too perfect. Beautiful archetypal Nordic woman. Perfect imagery. Short catchy songs. Immediately legible genre. Though no idea whether the lyrics make sense.
A longer conversation with ChatGPT later and, well, I let ChatGPT summarize the results:
VYRMA and Hyperpalatable Nordic Music
VYRMA is an interesting example of what might be called hyperpalatable Nordic music. The analogy is to highly engineered food: not necessarily bad, fake, or inferior, but unusually concentrated around qualities known to produce an immediate response. In VYRMA’s case those ingredients are easy to identify—dark female vocals, heavy ritual drums, minor and modal harmonies, repetition, ravens, wolves, snow, fire, braids, battle imagery and an atmosphere that signals “ancient North” within seconds. Tracks such as Sigr Í Norðri are remarkably catchy even when the listener understands none of the apparent language.
The project also has an unusually frictionless quality. Its visual identity is extraordinarily coherent, its central female figure looks almost archetypically Nordic, its catalogue appeared at a rapid pace, and the public presentation emphasizes the VYRMA identity rather than identifiable musicians. The official material describes VYRMA as a “digital music artist” or music project; a named female vocalist is difficult to find, while at least some official social-media material has carried an AI-content label indicating that sounds or visuals were altered or generated. What remains unclear is the exact division of labor: which vocals, instruments, images, compositions or lyrics are generated, and which are conventionally performed.
The language is perhaps the most curious part. VYRMA uses Old Norse imagery heavily, but its own wording tends to say “Old Norse inspired” rather than simply stating that all of the songs are sung in grammatical Old Norse. Authoritative lyrics and translations are surprisingly hard to find. Automatic transcription of Sigr Í Norðri produces little that is linguistically useful. An examination of the audio does show that the main refrain is not simply random vocalization: essentially the same phonetic and rhythmic sequence returns several times, without appearing to be a literal copy-and-paste of the same recording. There is therefore a stable “lyric” of some kind. What has not yet been established is whether that lyric is coherent Old Norse, imperfect reconstructed Norse, modern Scandinavian material, or stable Norse-sounding phonetic text whose main purpose is musical rather than semantic.
That ambiguity is part of what makes VYRMA interesting. Generative tools are exceptionally well suited to producing something that has all the perceptual signals of meaningful cultural material without necessarily preserving every layer underneath it. A listener who does not know Old Norse can recognize the sounds as “Nordic” without being able to test the grammar; similarly, a photorealistic singer can communicate an entire artistic persona without necessarily corresponding to a conventional performing artist. None of this requires deception. It may simply be a new kind of production in which the aesthetic identity comes first and the individual performer becomes optional.
The result can feel less like music that gradually developed from the peculiarities of particular musicians and more like an aesthetic target from which a musical project has been reverse-engineered. That helps explain the “hyperpalatable” quality: very little seems accidental. Each element reinforces the same response, and there is relatively little of the resistance, eccentricity or imperfection that often reveals the people behind a recording. Whether that is a virtue or a limitation is largely a matter of taste. But VYRMA is a useful glimpse of a larger change in music: AI does not merely allow people to imitate existing artists. It allows someone with a sufficiently clear idea—perhaps as simple as “I want dark, catchy Nordic music fronted by a striking blonde singer”—to construct the artist, the voice, the imagery and the catalogue around the idea itself.
Of course, ChatGPT can be wrong. I have not independently checked every source ChatGPT cited and my knowledge of music is limited. And as ChatGPT put it: «Some of this is inference rather than established fact. VYRMA does not appear to disclose clearly which parts of the project are conventionally performed and which are AI-generated.»
If this is a heavily AI-assisted project built by someone who wanted dark Nordic music fronted by a striking blonde singer, then kudos. It is professionally done. But perhaps too professionally — overdetermined. Human work comes with friction.
The strategic ambiguity about AI use does give me faint Milli Vanilli vibes. The audience is encouraged to attribute artistic performance and personal identity to the visible person while the actual causal chain may be materially different. I guess ChatGPT was right when it said «provenance becomes part of trust once the tool materially shaped the work».
And personally, once I began noticing signs of artificiality, I could no longer stop hearing them and eventually stopped listening.