How to Prompt Suno AI for Beginners: The GMIV Formula

Suno is not a conversational AI like ChatGPT. It does not want a chat. It needs the exact information to get from what is in your head to what actually comes out of your speakers. And the best way I have found to deliver that information, after 13 years of producing music and way too many hours inside Suno, is tags.

I call it tag based prompting, and it runs on a four part formula I named GMIV: Genre, Mood, Instruments, Vocals. Get those four right, in that order, and you are basically done. Everything else is decoration.

G is for Genre

Genre is arguably the most important part of the whole prompt, because Suno makes a bunch of assumptions based on the genre of your song. Rock has the built-in assumption that there's gonna be electric guitars, a drum kit, and a singer that's probably a little raspy. Say EDM and now it assumes it's loud and it's hype and it's energetic. So before you write anything else, the genre is already carrying half your prompt.

But there's a lot of nuance in genres, because under every main genre you have subgenres, and there are even subgenres of the subgenres. EDM has dubstep, house, techno, glitch hop, DnB. And dubstep alone has melodic dubstep, riddim, and a bunch more. So when I build a prompt, I start with the overarching genre first and then the subgenre: "Rock, Hard Rock" or "Rock, Grunge" or "EDM, Melodic Dubstep."

To show you how powerful this one word is, I generated the same exact prompt four times and only changed the genre: hard rock, emo rock, acoustic rock, thrash metal. One word changed per run. The hard rock and emo rock versions were completely different vibes. Thrash metal was a whole different universe. And the acoustic rock one still came out sounding like hard rock, so maybe Suno doesn't see acoustic rock as an actual genre. My bad. But even that teaches you something useful: Suno needs genre names it actually recognizes, so stick to labels that real scenes and real fans use.

M is for Mood

Mood basically breaks down into three things: tempo, emotions, and adjectives.

Tempo is the BPM of your track, and you write it as a number: "145 BPM." I've noticed that around 90% of the time, Suno will actually listen to the tempo instruction, which makes it one of the most reliable controls in the whole app. That's why I include a BPM in almost every generation I do. If you're not sure what tempo to use, just look up the general tempo range for the genre you're working in. Takes ten seconds and it can be a game changer for making sure your song has the right pacing.

Emotions are exactly what you think: happy, sad, angry, melancholic. You can get as deep as you want with these.

Adjectives are more about the overall vibe than an exact emotion: dark, brooding, intense, aggressive, chaotic. You can mix and match these and basically build your own mood from scratch. Tender and melancholic gives you one song, dark and chaotic gives you a totally different one, even if everything else stays the same. Use two to four of them and you've got a nearly infinite number of combinations.

I tested this one too: two identical instrumental indie rock prompts, same BPM, same instruments. One said "bright, hopeful, energetic" and the other said "melancholic, haunting, dark." Both came out as indie rock with the same instruments and the same jangly chord progressions, but they were completely different songs altogether. All from changing the mood.

I is for Instruments

Instruments are mostly self explanatory, but I've got two tips for you.

First, a really easy hack if you don't know what instruments to use: just look up what instruments are typically used in your genre. Making a 90s grunge song? Look up what's in one. There are literally thousands of instruments out there, and honestly it might be worth looking some up just to hear what they sound like. For the prompt itself, two or three instruments is the sweet spot. That gives Suno enough to work with without making it work too hard.

Second tip: put an adjective in front of the instrument. I generated four songs where the only change was the word before "acoustic guitar": nothing, aggressive, intricate, soft. The plain one was just some acoustic guitar. The aggressive one almost sounded like a metalcore song played on an acoustic. The intricate one had harmonics and way more notes going on. The soft one got gentle. Same exact instrument, four different feelings.

And if you keep experimenting like that, mixing different adjectives with different instruments, something cool happens: you eventually create your own sound, because you end up with your favorite words that nobody else is using in the same combination.

V is for Vocals

When you're doing vocals in your prompt, don't just write "female vocal." Who is she? Who is this woman singing? There are four elements you should include:

• Gender. Kind of self explanatory, but I felt like I should say it because it's there.
• Technique. I mean literal singing technique: falsetto, belting, head voice, chest voice. You can look up what techniques singers use in your genre, and then you can control exactly what the singer is doing.
• Emotion. Is she sad? Is she angry? Is she singing like she's perturbed by the void? You can go as deep as you want.
• Arrangement. Is it one lead singer, a choir, group shout vocals, background vocals, a duet? (Quick tip from experience: Suno's filter sometimes rejects the phrase "gang vocals," so write "group shout vocals" instead. Same sound, no rejection.)

Here's what that looks like in practice. "Acoustic pop ballad, female vocals" got me a singer that sounded pretty chill. Then I spiced it up to "female vocals, falsetto, desperate, emotional, intimate lead vocal" with the same lyrics. Full honesty: she didn't do the falsetto. That's Suno's fault, not mine. It doesn't obey every vocal instruction every time. But she did sound more desperate and emotional, and she went higher. The delivery changed in the direction I asked for, and that's what you're going for.

Putting it all together

So a GMIV prompt goes: genre and subgenre first, then your mood words with the BPM, then two or three instruments with their adjectives, then the vocal description. After that you can add more conversational stuff about the structure of the song, but the main gist is getting these tags correct. The right combination of tags basically is the prompt.

And if you want help with the vocabulary, that's literally why I made my two tools. The 10,000 SUNO Prompts eBook has Genre, Mood, Instrument, and Vocal tags mapped out for 200 genres, so you can look up the words for any sound instead of guessing. And SUNO GPT runs the whole GMIV formula for you: describe the song in your head in plain words and it builds the full prompt, BPM and all, ready to paste into Suno. The first 10 prompts are free, no card needed.