Inside Suno V6: Prompting Explained — Simple vs Custom, Styles vs Lyrics
Gary WhittakerInside Suno V6 · Part 5
Prompting Explained: Simple vs Custom, Styles vs Lyrics
Prompting in Suno V6 is not a contest to write the longest description. The useful skill is knowing how much control you need, putting each instruction in the right place, and changing one meaningful variable when the result misses.
Best for
Turning musical intent into usable direction.
Main job
Tell Suno what matters without making every field compete.
Watch for
Prompt problems that are actually model, structure or slider problems.
Next step
Edit the keeper instead of restarting everything.
Start with the right mode
Simple and Custom are two different jobs
Suno still gives you a fast description-first path and a more deliberate field-by-field path. In V6, Simple mode can accept surprisingly detailed direction, including richer instructions and multiple references. Custom mode is still the clearer choice when you want to separate lyrics, Styles, advanced settings and title into distinct controls.
Use Simple when the idea matters more than the setup
Describe what you want to hear and let Suno interpret the request. This is useful for quick concepts, mood-first creation and testing an idea before you need to manage every field.
Example: “A smoky late-night reggae soul song with deep male vocals, restrained drums and a chorus that opens into gospel harmonies.”
Use Custom when you need separation and diagnosis
Custom gives you dedicated space for your own lyrics, Styles and Advanced options. That separation matters once you need to know whether the problem is the writing, the global sound, a local section cue or a separate control.
Operator rule: if you need to diagnose the miss, Custom usually gives you cleaner evidence.
V6 changes the ceiling, not the principle. Suno says V6 understands more musician-language across vocals, instrumentation, structure, mood, references and feel. That means you can communicate more complex intent—but clarity still matters if you want repeatable testing.
The 20-second field map
What goes in Styles vs Lyrics?
| What you want to control | Put it here | Example |
|---|---|---|
| Genre / overall sound | Styles | Militant reggae rock, heavy bass, distorted guitar |
| Vocal character | Styles | Deep raspy male lead, urgent delivery |
| Mood / instrumentation | Styles | Defiant, cinematic, live drums, low strings |
| Words actually performed | Lyrics | Your verse, chorus, bridge and ad-libs |
| Song sections | Lyrics | [Verse], [Chorus], [Bridge] |
| One-section direction | Lyrics / local cue | [Final Chorus – choir responses, full intensity] |
| Something you do not want | Exclude / dedicated control | trap hi-hats, country twang |
| Length / influence / vocal-gender controls | Dedicated control when available | Duration, influence settings, Vocal Gender |
The goal is not a bigger prompt. It is a clearer job for every field. If Suno already gives the job its own setting, use that control instead of hiding the same instruction inside a paragraph.
Think in layers
Sound, structure and moments
1 · Sound
What should the whole song feel like? Genre foundation, production character, key instruments, broad vocal identity and emotional tone belong here.
2 · Structure
Where are the verse, chorus, bridge, drop or instrumental moments? Give the song an understandable architecture before adding micro-direction.
3 · Moments
What should change only here? A choir entering in the final chorus is a local instruction, not necessarily part of the entire song identity.
These directions are not deterministic code. They guide a generative model. The value of clean placement is that when the model misses, you have a better idea what to change.
Before and after
Fix placement before rewriting everything
Genre confusion
Weak:
reggae gospel trap cinematic rock emotional aggressive piano guitar choir 808
Stronger Styles:
Reggae-rock foundation, gospel choir accents, deep bass, distorted guitar, defiant cinematic energy
Choose a foundation. Make the rest accents or priorities.
Instructions get sung
If prose directions in Lyrics look performable, the model can treat them like material to sing. Move global musical direction to Styles and keep local cues concise.
[Bridge – instruments drop, close vocal] I hear the room breathe back my name
The chorus will not lift
If the global identity is already right, fix the moment instead of rewriting the whole Styles field.
[Chorus – full band, stacked responses, increased intensity]
Every retry changes too much
Lock what already works. Keep the same model and core brief, change one variable, listen, record the result, then decide the next move.
A better testing workflow
Stop treating every generation like a new experiment
Define one musical mission
Name the song’s job before you touch the fields: hook test, full-song candidate, mood study, instrumental bed or something else.
Choose Simple or Custom by the control you need
Use Simple for speed and whole-idea direction. Use Custom when you need field separation, your own lyrics or cleaner diagnosis.
Choose the model by the job
Keep V6, V6-Wild and V6-Mini as a separate decision from prompt wording. If you are testing a prompt, hold the model steady first.
Place each instruction where it belongs
Global sound in Styles, performed text and section cues in Lyrics, dedicated jobs in dedicated controls.
Generate a small comparable set
Do not rewrite the entire brief after one miss. Listen for which part of the instruction survived and which part drifted.
Change one meaningful variable
Prompt wording, model choice, structure cue and slider setting are different variables. If you change them all together, you learn almost nothing.
Diagnose before you regenerate
Prompt problem, model problem, slider problem—or song problem?
| Symptom | Check first | Do not immediately... |
|---|---|---|
| Polished result, wrong musical direction | Prompt mission and Styles hierarchy | Switch every setting at once |
| Direction is right but too safe | Model choice; compare V6-Wild | Stuff more adjectives into Styles |
| Direction changes wildly between tries | Model choice, prompt conflicts, Variety | Assume longer prompts equal more control |
| One section is weak | Local structure/edit workflow | Throw away the keeper |
| Reference is not followed closely enough | Relevant influence/reference control | Repeat the same reference request in multiple fields |
| Hook, cadence or lyric is generic | The writing itself | Blame the generator for a weak song idea |
Keep the useful existing training
Where to go when placement is no longer the problem
Your prompt is placed correctly but still misses
Move into a deliberate construction and repair process instead of cycling through random rewrites.
Build and repair the prompt →The song works but one section does not
Learn structure control before rebuilding a working song.
Learn Suno song structure tags free →You still need a clean starting framework
Build one prompt clearly before spending more credits or adding more controls.
Get the free AI Prompt Foundation Kit →You need examples by genre
Use the genre library as references, then adapt the prompt to your own musical mission instead of copying blindly.
Open the Suno prompt examples →Choose the bigger next step
Prompt placement is one skill. Build the system around it.
If this solved where your instructions belong, the next question is whether you need broader orientation or a repeatable AI-music operating system.
Need broader direction first?
The free Creator Starter Kit helps you place this lesson inside the larger creator journey: what you are making, how to think about rights and positioning, and what to do next.
Open the Creator Starter KitReady to build a repeatable music process?
Find Your Sound — AI Music Operator Manual V6 + Song Decision Guide connects prompting, structure, revision, production decisions and keeper selection into one reusable system.
Open Find Your Sound — Operator ManualPrefer to keep learning free? Continue with Find Your Sound — Free Core Path →
FAQ
Common prompting questions
Where do Suno prompts go?
Use Simple mode when one description can carry the idea. In Custom mode, put global musical identity in Styles, performed words and concise local section cues in Lyrics, and use dedicated controls for jobs Suno already separates.
Can Simple mode handle detailed V6 instructions?
Yes. Suno’s current V6 guidance says Simple mode can accept complex and diverse directions and can work with multiple kinds of references. Use Custom when you want clearer field separation and easier diagnosis.
Should instruments go in Lyrics?
Put overall instrumentation in Styles. A short Lyrics cue can make sense when an instrument matters only to one specific section, such as [Bridge – isolated cello].
Why does Suno sing my instructions?
Directions placed in Lyrics can resemble performable text. Move global production instructions into Styles and keep local cues concise and clearly separated from the lyric lines.
Should BPM and key go in Styles?
They can be part of the global direction when useful, but treat exact compliance as guidance rather than guaranteed deterministic control.
What if the prompt seems right but the result still fails?
Hold the model and core brief steady, isolate one variable and test it. The problem may be model selection, a structure cue, a slider, the source/reference, or the songwriting itself—not prompt wording.
Current platform references
What Suno currently confirms
Suno’s current V6 documentation says the model family understands more musician-language across vocals, instrumentation, structure, mood, references and overall feel. Its current Simple Mode documentation still centers on describing what you want to hear, while Custom Mode exposes separate lyrics, Styles and Advanced options. Interface details can change during rollout, so feature-specific controls should always be checked against Suno’s current documentation.
Series navigation
You have the direction. Next, protect the keeper.
Prompting gets you closer to the target. The next production skill is knowing when not to regenerate the whole song. Part 6 will focus on editing: local fixes, what to preserve, and when V6’s edit workflow is a better move than another full generation.
2件のコメント
Awesome thanks
It ist really amazing
Thanks