A knife slices through a bar of soap the size of a car. A thousand glass marbles fall in perfect slow motion. A block of kinetic sand gets cut by an oversized blade, one crisp crunch at a time. AI ASMR is one of the fastest-growing short-form video categories of 2026 — here's what's actually behind it.
AI ASMR takes the pre-existing human-filmed ASMR genre — soap cutting, slime, tapping, crushing — and pushes it somewhere a camera crew can't practically go: a candle the size of a building slowly melting, a thousand glass marbles falling at once, a block of kinetic sand the size of a car sliced by an oversized blade. Popular formats include kinetic sand cutting, glass-breaking, soap-cutting, mechanical-keyboard typing, candle-wax pouring, fruit-slicing, and ice-crushing. It's built on the same general-purpose AI video generators behind every other trend on this site, applied to a specific prompting style rather than a dedicated ASMR app.
What actually unlocked this category wasn't better visuals — it was models that can generate synced audio in the same pass as the video, rather than silent clips that need sound added in editing afterward. As of mid-2026, Veo, Kling, and Seedance can all produce native audio; earlier AI ASMR content leaned heavily on Google's Veo 3 specifically because it combined strong visual quality with genuinely convincing dialogue, sound effects, and music generated together. Kling's audio (particularly the version marketed as "Omni") is generally described as strong for ambient, general-purpose sound; some creators still reach for dedicated audio tools (like ElevenLabs) or dedicated ASMR-specific generators when they want sharper, more precise trigger sounds than a general video model reliably produces on the first try.
Standard per-clip pricing on whichever tool you use — audio generation is typically bundled into the base cost rather than charged separately, though higher-resolution/audio combinations can cost more credits per clip on some tools (see Kling's tiered audio pricing as one example). Full breakdown: AI video generator pricing compared.
Once you've got a clip you like
The video exports with its full generated scene (and audio) baked in — no built-in way to pull out just the subject. If you want to isolate a piece of the visual for a thumbnail, sticker, or a different composite, RemoveGifBG processes the MP4 frame by frame and returns a transparent result. See the AI video background remover guide.
Related: How AI videos are made · Tiny-world rescue trend · AI food-video trend · Pricing compared · Cheapest way to start
No single app — it's a prompting style (satisfying, repetitive physical interactions with an odd material, rendered at exaggerated scale) applied to general-purpose AI video generators that can produce native audio. Google's Veo, Kling, and Seedance can all generate synced sound as of mid-2026; earlier AI ASMR content leaned heavily on Veo 3 specifically because it was the first to combine strong visual quality with strong native audio.
Increasingly, the sound is generated natively by the video model itself in the same pass as the visuals — Veo, Kling, and Seedance all support this now. That said, several creators still layer in separately-generated audio from dedicated sound tools (like ElevenLabs) or dedicated ASMR generators when they want sharper, more precise trigger sounds than a general video model reliably produces.
Separate the prompt into clear parts — the visual action, the voice (if any), the specific sound effects, and any background music or ambience — rather than one run-on description. Focus on a single, satisfying, repetitive physical interaction with a distinctive material (cutting, crushing, tapping, pouring). Layering a low, quiet ambient drone underneath sharp foreground sounds (a knife through soap, marbles on glass) adds spatial depth that makes the close-up trigger sound more convincing.
A well-made AI ASMR clip is designed to loop seamlessly, and platforms' short-form algorithms read repeat views on the same clip as a strong engagement signal, pushing it to more viewers. That rewatch behavior — closer to how people replay a satisfying physical action than how they watch a normal video once through — is a real part of why this category grew so fast.