All Playbooksby edouardtheginger

AI ASMR: from prompt to post

Create one short video with several shots, add matching sound, edit it and schedule it. This example uses a 10-second clip. No n8n template or extra skill is needed.

The six parts

  1. Pick your ASMR concept.
  2. Generate one multi-shot clip.
  3. Add sound to the whole clip.
  4. Edit the finished clip.
  5. Connect your social accounts.
  6. Prepare captions and schedule.

Part 1: Pick your ASMR concept

GLASS STRAWBERRIES
Two strawberries made of red glass, one with a bite taken out of it.
CRYSTAL MANGO
Half a mango made of amber crystal, cut into cubes.
JEWEL GRAPES
A bunch of grapes made of purple faceted glass, one of them bitten.
GLASS HONEYCOMB
A piece of honeycomb made of golden glass, with broken shards beside it.

Step 1: Pick your concept

Choose glass strawberries, crystal mango, jewel grapes or glass honeycomb.

Step 2: Open ChatGPT or Claude

Copy the full prompt below into your chat. Replace the placeholders with your concept and audience, then send. Example: glass strawberries; crunchy ASMR fans.

Step 3: Save both outputs

The chat should return one VIDEO PROMPT and one AUDIO PROMPT. Keep both for the next two steps.

PROMPTpaste in ChatGPT or Claude
Turn [SELECTED CONCEPT] into one 10-second vertical AI ASMR mukbang clip for [TARGET AUDIENCE].

Return two labeled blocks:

VIDEO PROMPT: Write one Seedance 1.0 prompt under 1,500 characters. Define one adult fictional character, imaginary glass-like food, setting and lighting. Describe three brief shots within 10 seconds: reveal the food, show one complete bite, then show the satisfying texture and chewing. Keep the character and food consistent. Use 9:16 close-ups, simple camera movement and clear cuts. Start immediately. No injuries, dialogue, captions or logos.

AUDIO PROMPT: Write one MMAudio prompt for the entire clip. Describe close-miked bites, cracks, crunches and chewing specific to the material. Match the actual visible actions, including across cuts. Quiet background, controlled volume; no speech or music.

Keep the shots achievable in 10 seconds. Do not output separate generation prompts for each shot.

Part 2: Generate one multi-shot clip

Use one video prompt for the whole clip. Several shots share the selected duration; three shots do not mean three separate 10-second videos.

Generate the video

Seedance 1.0 · Lite / Pro / Pro Fast

FAL.AI
The fal logo.
WAVESPEED
The WaveSpeed logo.
KIE.AI
The Kie.ai logo.
REPLICATE
The Replicate logo.

Step 1: Choose one provider

Open a link below, sign in and add credits. WaveSpeed Lite 480p is the budget example.

Step 2: Set the generation

Choose text-to-video, 9:16 and 10 seconds. Use 480p for the budget test; choose 720p or 1080p where supported for a sharper result.

Step 3: Paste and download

Paste the VIDEO PROMPT from part 1 once. Generate, watch the entire result and save it as asmr-video.mp4.

Choose the model listed for your provider

fal.ai
Seedance 1.0 Pro Fast
WaveSpeed
Seedance V1 Lite T2V 480p
Kie.ai
V1 Pro Text To Video for multi-shot
Replicate
bytedance/seedance-1-lite

fal has retired its Lite endpoint and redirects to Pro Fast. Kie documents multi-shot support for Pro. Follow the linked provider page; variants and prices differ.

Budget example: one 10-second generation

WaveSpeed Lite video at 480p
$0.16
WaveSpeed MMAudio V2 sound
$0.01
Video + sound generation only
about $0.17

Published rate estimate, checked 6 October 2026. Excludes chat, editor and scheduler plans, credit top-ups, retries and higher-resolution output. Check the live generation quote before paying.

Part 3: Add sound to the whole clip

Generate the sound

MMAudio

FAL.AI
The fal logo.
WAVESPEED
The WaveSpeed logo.
REPLICATE
The Replicate logo.
EACHLABS
The Eachlabs logo.

Step 1: Choose MMAudio

Open one of the four providers below. fal and WaveSpeed offer MMAudio V2; the other listings are MMAudio.

Step 2: Upload your video

Upload asmr-video.mp4. Paste the AUDIO PROMPT from part 1. Match the sound duration to the actual clip, usually 10 seconds here.

Step 3: Listen and save

Check every bite and cut with headphones. Download the video with sound as asmr-sound.mp4, or keep the separate audio beside its video.

Use your generated AUDIO PROMPT first. This reusable version is useful when you need to revise the sound after seeing the actual clip.

PROMPTpaste in MMAudio
Close-miked ASMR of [FOOD OR MATERIAL] being [VISIBLE EATING ACTION].

[SOUND TEXTURES FOR THE CLIP].

Match each bite, crack, crunch, and chewing sound to the corresponding movement in the uploaded video. Follow the actual action timing across all shots, with delicate texture sounds between stronger crunches.

Use a quiet background, clear detail, and controlled volume. No music, speech, narration, unrelated effects, harsh distortion, or exaggerated reverb.

If the sound misses a bite: describe the visible action more precisely and rerun the audio only. If your download contains audio alone, align it to the video start in the editor and keep the tracks linked.

Part 4: Edit the finished clip

Edit your video

AI video editors

CHATCUT
The ChatCut logo.
DESCRIPT
The Descript logo.
KAPWING
The Kapwing logo.
VEED
The VEED logo.

Step 1: Import your clip

Open one editor below and upload asmr-sound.mp4. If sound is separate, import both files and align their starts.

Step 2: Open its assistant

Use ChatCut chat, Descript Underlord, Kapwing’s video assistant or VEED’s assistant. Paste the prompt below and identify the uploaded clip.

Step 3: Review and export

Check the cuts and audio sync. Make any unsupported adjustment in the editor, then export the final MP4.

PROMPTpaste in your AI editor
Edit [VIDEO CLIP] into a polished vertical 9:16 ASMR video.

Keep the clip’s audio linked and synchronized when trimming. Remove black frames, frozen endings and unnecessary waiting before or after the action. Preserve complete bites, satisfying crunches and quiet texture sounds.

Use clean cuts and consistent volume. Prevent clipping and abrupt audio clicks. Do not add music, voiceover, captions, transitions or generated footage. Keep the original playback speed.

Keep the natural duration without padding. Preserve the source resolution and export an MP4.

Flag any remaining visual artifacts or synchronization problems for review.

Final playback check: watch with headphones. Keep the complete bite, remove dead time and check for distorted faces, broken food geometry, black frames or an abrupt ending. Export at the source resolution; enlarging 480p does not restore detail.

Part 5: Connect your social accounts

Schedule your video

Connect your social accounts

POSTIZ
The Postiz logo.
UPLOAD-POST
The Upload-Post logo.
BUFFER
The Buffer logo.
AYRSHARE
The Ayrshare logo.

Step 1: Pick one scheduler

Postiz is the example here. Create an account and connect the social profiles you want to use through its dashboard.

Step 2: Connect your AI agent

Add the official Postiz connector in ChatGPT or Claude where available, or follow its MCP setup guide below. Sign in to your Postiz account.

Step 3: Check the connection

Ask: “List my connected social accounts.” Confirm the returned profile names before giving the agent your video.

MCP is a connection that lets your assistant use an app. A prompt alone cannot connect accounts or schedule posts. Use one supported connector; you do not need all four services.

Connection guides

Postiz
Official connector or hosted MCP; browser sign-in.
Upload-Post
Hosted MCP; browser sign-in or API-key setup.
Buffer
Use its integration guide for your AI tool.
Ayrshare
Action MCP with an API key; not the docs-only server.

If a connector requires an API key, enter it in the connector’s secure settings. If your agent cannot upload local files, use the scheduler’s upload area and provide the resulting accessible media link.

Part 6: Prepare captions and schedule

From video to socials

01 Finished video
Give your agent the final MP4.
02 Prepare captions
Write a caption for each platform.
03 Review + approve
Check captions, accounts and times.
04 Schedule the posts
Use your connected publishing app.

Confirm each post in the scheduled queue.

Step 1: Give the agent the video

Attach your final MP4 or provide a supported media URL. Replace every placeholder below, including the time zone.

Step 2: Review the plan

Check the exact accounts, captions, date and time. Keep the AI-generated disclosure. Approve only the version you want posted.

Step 3: Confirm the queue

Check the scheduler for one post per destination. A message saying “done” is not enough; verify its status and scheduled time.

PROMPTpaste in your connected agent
Prepare [FINISHED VIDEO FILE OR URL] for publication using [CONNECTED PUBLISHING SERVICE].

Video concept: [CONCEPT].
Destination accounts: [PLATFORMS AND ACCOUNTS].
Schedule: [DATE, TIME, AND TIME ZONE].

Write a short caption for each platform with relevant hashtags. Describe the video as AI-generated ASMR. Avoid earnings promises, invented claims, or presenting the footage as real.

Check each platform’s upload requirements and apply supported AI-content disclosures.

Show me the captions, destination accounts, and posting times for approval. After approval, schedule each post once.

Return each platform’s confirmed scheduled time and post ID or link. Report failed or unsupported actions clearly. Do not claim a post is scheduled without confirmation from the publishing service.

Keep your files: save the video prompt, audio prompt, final MP4 and editing project together. If a post fails, check its status before retrying so you do not create duplicates.