AI ASMR videos: how to make them, where the sound comes from, and what each one costs

Hook Studio is an AI reel maker that turns a chat into short vertical videos for TikTok and Instagram. For AI ASMR videos it does one half of the job: the picture. The clips it makes come back silent, or with speech. The crunch, the tap, and the slice come from a sound you pick in the TikTok or Instagram library when you post.

That is a real difference from the tools that made this trend famous, so this guide says it up front, then shows what you can do about it. Prices and behavior are as of October 26, 2026.

The wave started in mid-2025, when Google's Veo 3 began producing video and sound in one step. Coverage at the time described AI-made ASMR clips on TikTok with millions of views, mostly glass fruit being sliced and people biting into molten rock.

The sound was the point. Earlier generators made silent video, so you had to find or record a crunch and line it up by hand. A model that generated the slice and its sound together turned a two-step job into one. One venture capitalist called native sound a huge unlock.

What the trend looks like now

The subjects fall into a few families. The one you will find most is AI glass fruit cutting: a knife through a translucent strawberry, orange, or watermelon. Lava and molten-rock clips are the second. Crystals, kinetic sand, and soap or slime pressed into shapes follow the same logic. They share one trick: a surface that does something a real object could not.

Reception is split. Some viewers find it soothing. At least one tech outlet said it left the writer more confused than relaxed. Both reactions keep people watching, but neither is a promise that your clip will do well. If you want the fruit version in depth, read our guide to AI fruit videos, and for the wider picture see viral AI video trends.

Hook Studio makes the picture and the motion. It does not generate ambient sound, music, or foley. Here is the line, so you can decide before you spend a credit.

• It does: make a still of the scene, make a 4 to 10 second clip from it, add on-screen text, join shots, and export a 9:16 reel.
• It does not: generate the crunch, the tap, or any room sound. Clips come back silent, or with speech if you ask for a voice.
• How the sound gets in: you pick a sound from the TikTok or Instagram library when you post, or you upload your own audio to the reel, or the agent adds a 5 credit voice line.

If your whole idea is that the generated sound has to match the visual exactly, a model with native audio is the better fit. If your idea is the visual, and a library sound does the rest, this works well and costs less.

Make a still first, then a clip from the still you like. A still costs 11 credits and a clip costs between 34 and 96, so you want to be sure of the picture before you pay for motion.

• 1. Describe the scene in one line. For example: a clear glass strawberry on a white marble counter, a steel knife just touching the top.
• 2. Steer the still. Say what to change in a short line: "cleaner background", "closer", "more see-through". Each edit keeps the scene and costs 11.
• 3. Make the clip. Pick the model under the message box. Ask for one slow action, like the knife sliding down.
• 4. Add text if you want it. A short caption, nothing on the object. Then render the reel for a few credits.
• 5. Download and post. Pick the sound inside TikTok or Instagram.

One finished 5 second clip costs 45 to 107 credits, depending on the model. That is about $0.39 to $0.93 on the Concept plan. Pay as you go, your first $5 costs $3 (500 credits), which covers several of these.

These prices leave out the automatic realism checks, a few credits each, and any extra still edits. A failed clip refunds its credits, and the agent shows each price before anything runs. If you want the full picture of how clips are priced, see Kling AI in Hook Studio.

Which one to pick

• Seedance 2 Mini for a loop or a test. A slow slice on a clean surface does not need a big model, and 45 credits lets you try several ideas.
• Kling 3 for the one you will post. It makes 1080p, and glass, light, and liquid read better at that size.
• Try the cheap one first. If the motion looks right on Mini, you know the idea works before you pay for the sharper version. The final on Kling 3 will not match the Mini take exactly, so treat Mini as a test of the idea, not a preview.

A satisfying clip is a close-up of one slow action on a clean surface. Nothing in the frame competes with it, and the action finishes. That is a design rule more than a style, and it works the same with or without AI.

• Close-up. Fill the frame with the object. A small object in a big room reads as a scene, not as a texture.
• Slow. Ask for slow motion or a steady pace. Fast cuts break the calm.
• One action. One cut, one press, one pour. If two things happen, the eye does not settle.
• Clean surface. A plain counter, one light, no clutter. Anything extra can also make the model draw something strange.
• A clean ending. The clip should land on the cut finishing, or loop back to the start.

This is also why it is cheap to try. A 5 second clip is enough for one action, and the video does not need a face, so there is no avatar to keep consistent. If you want to put this clip inside a longer reel, our AI reel maker page shows how the other shots are built.

The sound goes on last, and the easiest place to add it is inside TikTok or Instagram. Download the reel from Hook Studio, start a new post, and choose a sound from the library. Direct posting from Hook Studio is not available yet, so you post from the app.

• Library sound (recommended). Search for slicing, crunch, or ASMR sounds in the app. The platform handles the license. On TikTok, business accounts can use only the Commercial Music Library.
• Your own audio. Record a real crunch or tap and replace the audio in Hook Studio. It is a free edit. Matching a real sound to an AI slice takes a few tries, so trim until the hit lands on the cut.
• A voice line. If your video has a whispered intro, a voice line costs 5 credits. It is one stock voice, and Hook Studio does not clone voices.

Yes: TikTok and Instagram ask you to label realistic AI-generated content, and a glass strawberry that looks like a photo is exactly the kind of clip that confuses people. Every Hook Studio download comes with a reminder to turn on the AI label. Switch on the platform's own AI label when you post as well.

Hook Studio is not affiliated with TikTok, Instagram, Meta, Google, or Kuaishou (the company behind Kling). Nothing here promises views or followers. It describes what tends to hold attention and what it costs.