←Back/Marketing/GPT-4o
IntermediateMarketing

How to Create an AI-Generated Music Video with GPT-4o and Hedra

Create a short music video by defining a visual treatment, generating and refining stills, animating permitted audio in Hedra, and editing multiple brief clips into a coherent sequence with documented media rights and synthetic disclosure.

How to Create an AI-Generated Music Video with GPT-4o and Hedra

Source episode · 06:00

How I AI: Anish Acharya's 3 Creative AI Workflows for Music Videos, Book Cataloging, and Personal Finance

with Claire Vo

Anish creates a still image with GPT-4o, brings it into Hedra with a prepared vocal track for lip sync, optionally separates vocals with Demucs, iterates from a clean concert look toward a grimier 1990s Seattle aesthetic, and assembles short generated clips into a music video.

Read the full episodeWatch on YouTubeListen on SpotifyListen on Apple

Before you start

What you need

  • Audio you created, licensed, or otherwise have permission to use
  • A visual brief with story, era, setting, wardrobe, camera language, and exclusions
  • Rights cleared reference material and a policy for real person likenesses
  • GPT-4o image generation, Hedra, optional Demucs, and an editor such as Kapwing
  • A shot list, target aspect ratio, clip duration, export settings, and disclosure plan

What you’ll make

A short edited music video with consistent visual direction, usable lip sync, clean audio, recorded prompts and provenance, and the permissions and disclosure needed for its intended audience.

Tools used

  • GPT-4o

    OpenAI's multimodal model

Step by step

The workflow

Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.

5 steps

Step01

Generate the Main Image with GPT-4o

Write the treatment before generating frames. Define the fictional or authorized performer, setting, era, wardrobe, emotional beat, composition, camera treatment, aspect ratio, and features to exclude. Use visual properties instead of asking for a copy of a living artist or protected video.

Example prompt
Create a still frame for a synthetic music video. Subject: [fictional or authorized performer and action]. Setting and era: [details]. Wardrobe: [details]. Camera and image treatment: [composition, movement cue, lens, texture, color]. Emotional beat: [beat]. Aspect ratio: [ratio]. Do not resemble a named real person, include logos, or reproduce a specific copyrighted frame.
Step02

Animate the Image with Hedra

Upload the selected still and a permitted audio clip to Hedra. Keep the first animation short, specify restrained movement and camera behavior, and test facial stability and sync before generating a full sequence.

Example prompt
Animate this image for [duration] using the supplied authorized audio. Keep identity, wardrobe, background, and lighting stable. Motion: [performance and camera cue]. Prioritize natural mouth movement and subtle expression. Avoid extra people, text, logos, face changes, and sudden camera jumps.
Step03

Source and Prepare Audio

Prepare the audio from an original or licensed source. Trim the exact phrase, normalize levels, and preserve a record of ownership, license, source file, and edits before uploading it to any model.

Step04

Isolate Vocals with Demucs (Optional)

If permitted and necessary, use Demucs to isolate vocals or instruments. Inspect bleed and phase artifacts, retain the original mix, and do not treat stem separation as permission to reuse a copyrighted recording.

Example prompt
demucs two-stems vocals /path/to/audio.mp3
Step05

Assemble the Final Video in Kapwing

Generate a shot list of short clips, edit them in Kapwing or another editor, and use cuts to create continuity and hide generation limits. Check sync, pacing, artifacts, titles, audio levels, credits, rights, and disclosure before export.

Example prompt
A short video clip with a [1990s Seattle grunge] aesthetic. The footage should look like it was shot on a [grimy camcorder]. Show [an empty high school band auditorium with dystopian energy].

What good looks like

  • Every audio, image, logo, likeness, and reference has documented permission or an acceptable use basis.
  • The generated performer is fictional or authorized and is not presented as a real recording.
  • Short clips maintain intentional continuity of wardrobe, setting, camera treatment, and emotional arc.
  • The final export has intelligible synchronized audio, no obvious visual artifacts, and appropriate synthetic media disclosure.

Build your next product with ChatPRD

Turn an idea into a PRD, user stories, and a plan.

Try ChatPRD free

After the steps

Runbook notes

How to recover when the loop fails and where human judgment helps.

Recover

If it goes sideways

The workflow uses a commercial recording, recognizable performer, or trademark without permission
Use original or licensed audio and fictional or authorized subjects, and obtain rights guidance before public or commercial release.
A generated frame closely resembles a real artist or implies an event that occurred
Remove named likeness requests, redesign the performer, and label the work as synthetic where viewers could be misled.
Lip sync drifts or the face deforms during longer phrases
Use a clean short vocal segment, a front facing source image, shorter shots, and cut away before artifacts accumulate.
Individual clips look attractive but do not belong in the same video
Lock the treatment, recurring subject details, palette, lens and camera cues, and change one visual variable at a time.

Keep building

Related workflows

Create an Automated Workout Tracker with a Simple Claude Voice Note
beginnerPersonalClaude

Create an Automated Workout Tracker with a Simple Claude Voice Note

Log your workouts effortlessly without ever opening a spreadsheet. This workflow uses the Claude mobile app to parse a simple voice note and automatically update your workout tracking spreadsheet in the background.

Open
How to Rebuild Your Gmail Inbox Inside Claude to Manage Email
advancedOperationsClaude

How to Rebuild Your Gmail Inbox Inside Claude to Manage Email

Create a custom email management interface within Claude to read, triage, and draft replies without opening the Gmail tab. This workflow helps you reclaim focus and compound your learnings by training your AI on every interaction.

Open
Automate Client Proposals and Onboarding with a Claude 'Pipeline Operator'
intermediateSalesClaude

Automate Client Proposals and Onboarding with a Claude 'Pipeline Operator'

Build a fully automated system in Claude to generate personalized, interactive HTML proposals and client onboarding materials, saving hours on administrative work and creating a high-touch experience.

Open

Start shipping
better products.

Join 100,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.

Start building freeRequest a demo
Free to start
No credit card
SOC 2 certified
Enterprise ready