After this lesson you'll know
- How to write effective prompts that produce the sound in your head
- The anatomy of a good AI music prompt (tempo, genre, mood, instruments)
- How to iterate and refine outputs until they hit the mark
- Basic song structure and how to guide AI toward professional arrangements
Open Suno. Type a sentence. Get a track. It really is that fast.
Let's do this right now. Go to suno.com and sign up (free tier gives you 50 credits/day). Click "Create." In the prompt box, type:
"Chill lo-fi hip-hop beat, dusty vinyl texture, mellow piano chords, soft boom-bap drums, 85 BPM, rainy day mood"
Hit generate. In about 30 seconds, you'll have two variations of a complete instrumental. That's it. You just made your first beat. Now let's learn how to make it actually good.
The difference between a random generation and a track worth releasing comes down to one skill: prompt engineering for music. And it's a skill you can learn in an afternoon.
Every killer prompt has these five elements.
Think of a music prompt like a creative brief to a session musician. The more specific you are, the better the output. Here's the formula:
1. Genre + Subgenre: "Hip-hop" is vague. "Memphis trap with Three 6 Mafia influence" is specific. The more precise your genre reference, the more coherent the output. Suno and Udio both understand hundreds of subgenres.
2. Tempo (BPM): This controls the energy. 70-90 BPM for chill/lo-fi. 90-110 for R&B/soul. 120-130 for pop/house. 140-160 for drum & bass/trap. 160-180 for punk/hardcore. Always include BPM — it prevents the AI from guessing wrong.
3. Instruments & Sounds: Name them. "Electric guitar, Rhodes piano, 808 bass, hi-hats, claps" is 10x better than "a beat." Include textures: "vinyl crackle," "tape hiss," "reverb-drenched," "lo-fi warble."
4. Mood & Emotion: "Melancholic sunset drive" produces different results than "aggressive gym energy." AI models are trained on millions of descriptions — emotional language works surprisingly well.
5. Structure Keywords: "Verse-chorus-verse-chorus-bridge-chorus" tells the AI your arrangement. You can also use: "build-up to drop," "gradual intensity increase," "ambient intro into hard beat," or "stripped-back breakdown at 1:30."
Your first generation is a starting point, not a finished product.
Professional AI producers never ship the first output. They iterate. Here's the loop:
Generate → Listen → Identify what works → Adjust prompt → Regenerate.
Listen to your first output and ask: Is the tempo right? Is the mood right? Are the instruments what I envisioned? Is the structure coherent? Write down what you'd change. Then modify your prompt and generate again.
Common adjustments:
Too busy? Add "minimal" or "sparse arrangement" to your prompt. Too generic? Add more specific references ("sounds like Tame Impala meets Boards of Canada"). Wrong energy? Adjust the mood descriptors and BPM. Vocals when you want instrumental? Add "[Instrumental]" tag in Suno or specify "no vocals, instrumental only."
Typically, 3-5 iterations gets you from "interesting idea" to "this actually slaps." That's still only 15-20 minutes of total work. Compare that to the 4-8 hours a traditional beat takes from scratch.
You don't need music theory. You need to understand sections.
Every song is just sections arranged in order. Here are the building blocks:
Intro (4-8 bars): Sets the mood. Usually a stripped-down version of the main beat. Think of it as the "loading screen" for the listener's brain.
Verse (8-16 bars): The storytelling section. Usually less intense than the chorus. If you're making beats for rappers, this is where they'll spit bars — keep it spacious.
Chorus/Hook (4-8 bars): The most memorable part. Higher energy, more layers, the part people sing in the shower. This is where your beat should hit hardest.
Bridge (4-8 bars): A departure. Different chord progression, different energy. Creates contrast that makes the final chorus land harder.
Outro (4-8 bars): Wind it down. Strip elements away gradually. Some producers just reverse the intro.
Common structures: Pop = Intro-Verse-Chorus-Verse-Chorus-Bridge-Chorus-Outro. Hip-hop = Intro-Verse-Hook-Verse-Hook-Verse-Hook-Outro. Electronic = Intro-Build-Drop-Breakdown-Build-Drop-Outro. Include these patterns in your prompts for more professional arrangements.
Lock it in.
Quiz
1What are the five elements of a killer AI music prompt?
2How many iterations does it typically take to go from first generation to release-worthy?