HOW ZELQAVIN WORKS
ZELQAVIN
breathes.
The rest runs on electricity.
My music begins with my experiences and ideas.
Generative AI plays a central role in developing those ideas into lyrics, music, and vocals.
The tools and methods may change.
My involvement does not mean that I personally write, perform, or construct every part.
HUMAN DIRECTION
Every work begins with something I want to explore.
I describe the direction, discuss possibilities with AI, and review what develops.
I try to leave out as many details of my personal circumstances as I can, so a listener can feel like the person at the centre of a song.
Some personal details remain.
I aim for low, matte vocals and restrained expression, while allowing the sound to be forceful, rhythmic, or rough.
The result does not always fully match the intention.
GENERATIVE AI
I use generative AI extensively for lyric writing and translation, music prompts, music generation, and vocals.
Its contribution is substantial.
Lyrics can develop through conversations about my experiences and ideas.
AI writes much of the wording; I provide direction, review the results, remove material, and make revisions where needed.
Reviewing a lyric does not mean I wrote every line.
AI also helps turn the musical direction I describe into generation prompts.
I listen to the results and request further versions or revised prompts when needed.
My own voice is used as a reference in the AI vocal generation process.
The finished singing is AI-generated, rather than a direct recording of me performing the complete song.
SELECTION IS PART OF CREATION
A single idea can lead to many possible outcomes.
Sometimes I keep a result, sometimes I request another version, and sometimes I discard it.
AI’s suggestions are part of that process.
I may follow them or disagree.
Selection includes deciding what to keep, what to leave out, and when to accept a suggestion.
My attachment to the experiences behind a song can influence those decisions.
WORKING PRINCIPLES
There are artists and songs I genuinely admire.
Respect for their work is one reason I set boundaries for how I create music.
I do not ask generative tools to recreate a particular artist, a released song, or a specific singer’s voice.
I do not use other artists’ commercially released recordings as source material for music or vocal generation.
Musical styles and shared conventions can be explored without aiming to reproduce another work’s distinctive expression.
When a work draws on personal experience, I avoid presenting imagined incidents or dialogue as remembered facts.
I make reasonable efforts to check for potential similarities and rights-related risks before release.
If concerns remain, I may revise the work or choose not to release it.
No review can guarantee that a work is free of resemblance or legal risk.
ONE WORK, MANY LAYERS
Lyrics, sound, atmosphere, and narrative can belong to the same creative world.
Continuity does not require every work to repeat the same subject, rhythm, or texture.
My work can also include sounds I record, audio editing, and finishing, with AI providing guidance as well as generated material.
These contributions are different from writing every lyric or performing every musical part.
WHAT REMAINS
I care about how the work is made, and I want that process to be described accurately.
The background explains where a song began; it does not settle what it must mean.
The tools used and their contributions can vary from release to release.
Specific production credits and tool roles are recorded in the history of each album or single.