Lo-fi has always been a forgiving genre. It tolerates rough edges. It rewards mood over precision. It thrives in long loops, simple harmonic ideas, and producers willing to leave the room before the track has technically finished. For a long time, that forgivingness was a virtue — it let bedroom producers in São Paulo and Seoul make records that lived for years on the same playlists as artists with million-dollar studios.
In 2026, that forgivingness is also the genre's biggest liability.
Lo-fi is the easiest sound in the world to generate with AI. A prompt of fourteen words gets you a tolerable Rhodes-and-drums loop in twenty seconds. Anyone can ship a thirty-track EP by Friday. Spotify processes more lo-fi than any other genre, and most of it is wallpaper.
The bar for what counts as a real lo-fi record has gone up. Not because AI is bad. Because the floor moved.
What changes
The economics. Producing a track no longer takes a weekend. It takes an hour, including the cover. The cost per release has collapsed to almost nothing. This means that the labels and producers who survive will be the ones who have something to say beyond "I can make a lo-fi track" — because everyone can make a lo-fi track.
The signal. Listeners can sniff out generic faster than ever, because they're drowning in it. Save rates, skip rates, and playlist conversion all reward records that have a recognisable point of view. Anonymous music has never been less competitive.
The expectations of supervisors. Sync agencies and music supervisors have started filtering AI-flooded catalogues by curatorial taste, not raw volume. A label of eighty thoughtful releases now matters more than a label of eight hundred undifferentiated ones.
What stays
Mood is still mood. The lo-fi tracks that have lived on playlists for five years are the ones that nailed a specific feeling — late autumn café, third hour of studying, gentle rain on a Tuesday. That hasn't changed. It can't be changed.
Restraint. A two-minute track that does one thing well will always beat a four-minute track that does six things adequately. Most AI-generated lo-fi suffers from a lack of editing — the producer never stepped in to delete the unnecessary parts.
Sequencing. A great lo-fi record is built like a Sunday afternoon. The order of tracks matters. The transitions matter. The decision to leave silence between songs matters. None of this is automated.
Cover work. A cover image that feels like the music is doing half the work of selling it. Most AI-generated lo-fi has a generic Tokyo-street or 80s-anime cover. That look is dead. We're betting on intimate, restrained, photographic.
How we work on Velour Tape
Our lo-fi project is built on a simple workflow. Forty prompts a month, generated in batch. Of those forty, twelve get extended into full tracks. Of those twelve, four become releases. Of those four, the ones that get traction on playlists get a second life as remixes, instrumentals, and 30-second loops.
Every prompt is anchored in a specific mood: late-night journaling, autumn coffee, rainy window. Every release is paired with a photograph-style cover, never an illustration. Every track is mixed for headphone intimacy, never for car speakers. And every release is logged in our internal sheet with BPM, key, tags, save rate after week one, and the prompt that generated it — so the catalogue learns from itself.
The result, we hope, is a small, recognisable body of work. Not the most lo-fi anyone makes this year. Just the most consistent.
The genre will be fine
Lo-fi as a sound is too useful to die. Students still need to study. Designers still need background music. People still want a sonic blanket for the hours between things. The category will keep growing.
What will shrink is the middle. Generic lo-fi will get cheaper to produce and harder to monetise. Distinctive lo-fi — with a voice, a point of view, a recognisable mood — will get more valuable than ever.
That's the bet we're making with Velour Tape, and with Mossfield. It feels like the right one.
— Velour Tape · Lisbon, May 2026