AI-Powered Educational Music Across All Genres
Purpose
IB Music Production is our internal workflow for turning IB concepts into genre-flexible educational music.
Students learn best when content aligns with their personal tastes — so our goal is to create multiple musical interpretations of each topic, transforming complex concepts into catchy, memorable songs supported by high-impact visuals.
This workflow uses ChatGPT (lyrics) → Mureka AI (melody + vocals) → OpenArt (music video) to produce a complete learning asset.
Overview of the Workflow
1. Generate Lyrics (ChatGPT)
We begin by producing clear, memorable, syllabus-accurate lyrics for a chosen IB topic.
Key steps:
•
Specify the length (e.g., 2–3 minutes).
•
Specify the genre (e.g., pop, K-pop, rap, hip-hop, ballad, country, and more).
•
Provide the topic (e.g., Organic Chemistry — Isomerism).
•
Ask ChatGPT to rewrite technical terms into phonetically readable form:
◦
“p-K-A” instead of “pKa”
◦
“E-N-thal-pee” instead of “enthalpy”
Why this matters:
Most music generators struggle with raw scientific jargon unless spelled out phonetically.
2. Generate Audio (Mureka AI)
Mureka AI generates the instrumental, melody, vocals, and genre style using the lyrics.
Steps:
•
Paste the lyrics.
•
Select the genre (pop / R&B / rap / lo-fi / country / EDM / K-pop / ballad).
•
Provide musical references (artists, groups, eras).
•
Choose the voice type (male / female / youthful / deep / soft).
•
Select the model version:
◦
Model 01 (5 credits) for highest quality
◦
1-credit models for drafts
Output:
•
2–3 minute track exported as MP3.
Notes:
•
Our current plan supports ~70 full-quality songs.
•
Editors should create multiple genre versions of the same topic to support different student preferences.
3. Generate Music Video (OpenArt)
OpenArt creates the visual layer — a narrative music video synced to the Mureka audio.
Steps:
1.
Build → Music Video → Narrative Mode.
2.
Upload the MP3 from Mureka.
3.
Select a character and style (can vary by genre).
4.
Write a ≤500-word storyline describing the concept visually.
5.
Generate storyboard.
6.
Fix inaccuracies:
•
OpenArt has no chemistry understanding
•
Scenes may look visually appealing but scientifically wrong
•
Editors must regenerate each scene using precise, explicit ChatGPT-written prompts
7.
Regenerate images → convert to video per scene.
8.
Download the final MP4 (audio included).
Cost:
•
A single 3-minute video uses ~11,000 credits (estimate).
•
Editors must optimise retries by improving prompt quality.
Quality Requirements & Challenges
Accuracy:
•
Molecules, orbitals, diagrams, lab setups must be corrected manually.
•
Simplify visuals for general audiences while retaining conceptual truth.
Prompt Engineering:
•
Longer, more detailed prompts = better results.
•
Avoid scientific shortcuts — assume the AI knows nothing about chemistry or physics.
Genre Variation:
•
Each topic should eventually have multiple genre versions to maximize student engagement.
•
Editors must understand how musical genre influences lyric flow and visual mood.