The Expert Secret Sauce Guide to Suno v6 prompting: Production Notes and Style Tags

1. What you need

• A Suno Pro or Premier account with credits for six generation requests.
• Suno’s browser interface, with the core v6 model available.
• Custom mode, with Instrumental switched off.
• Headphones and a fixed playback volume for comparisons.
• A plain-text editor and a folder named night-window-v6.
• The original example lyrics below, or replacement lyrics you control.

These steps target core v6, not v6-wild or v6-mini. Suno lists core v6 under Pro and Premier subscriptions. Different variants require separate listening tests.

2. Step 1 — Set up your Suno v6 prompting baseline

Open Suno’s Create page and select Custom. Choose v6 from the model dropdown near the Create form’s upper-right corner.

Switch Instrumental off. Remove attached audio references, selected Personas, and other optional reference inputs for this exercise. These inputs would introduce additional variables.

Open the advanced controls and enter these starting values.

| Control | Starting value | | --- | --- | | Weirdness | 20% | | Style Influence | 75% | | Variety | 0, or Off | | Exclude / Exclude Styles | Leave empty initially |

These values are experimental starting points, not official optimal settings. Weirdness runs from Safe to Chaos. Suno identifies 50% as its normal expected result.

Style Influence controls adherence to the style input. Variety can adjust that input, so keep it at zero during testing.

Create session-notes.txt inside your project folder. Record the model, three control values, and today’s date. If your interface lacks a listed control, record that difference before generating.

Your screen should show core v6, editable lyrics, and a separate style field.

3. Step 2 — Write an audible production brief

In session-notes.txt, add the following brief. This is your listening checklist, not another Suno input field.

Project: Night Window.
Genre: Electronic synth-pop.
Tempo target: 104 BPM.
Meter target: 4/4.
Verse: Sparse drums, muted synth pulse, centered lead vocal.
Chorus: Stronger drums, wider synth chords, backing harmonies.
Bridge: Remove drums, retain pad and lead vocal.
Ending: Final chord decay, without an abrupt cutoff.

Treat tempo and meter as requested musical characteristics. Text instructions do not establish a locked timing grid.

Add three priorities beneath the brief.

1. The chorus must sound fuller than the verse.
2. The lead vocal must remain understandable.
3. The bridge should remove the drum pattern.

For Suno v6 prompting, these priorities determine which draft you keep. They also prevent attractive but unrelated details from dominating your decision.

Avoid requests such as “professional,” “perfect,” or “radio-ready.” Replace them with audible behavior, such as “short vocal reverb” or “restrained cymbals.”

Your saved brief should describe changes you can identify during playback.

4. Step 3 — Enter style tags as production notes

Find the Styles field. Some interface variants label this Style or Style of Music.

Paste this complete block into that field.

Electronic synth-pop at 104 BPM in 4/4.
A muted synth pulse anchors the verses.
Rounded synth bass supports a tight kick and restrained snare.
A warm alto lead sings clearly, with restrained vibrato.
Keep the lead centered, with short reverb.
Make choruses fuller through wide synth chords and soft backing harmonies.
Drop the drums during the bridge.
End with a final chord decay.

Keep the line breaks if the field accepts them. The musical instructions remain readable without them.

Here, “production notes” means ordinary descriptive language. It does not mean a verified command system or hidden control panel.

The first line establishes genre, tempo, and meter. Later lines assign instrument roles, vocal delivery, and arrangement changes.

Do not append artist names or unrelated genres. Do not add exact compressor ratios, EQ frequencies, or loudness targets. Those requests would not provide verifiable engineering control over the generated mix.

Save this block as style-v1.txt. Suno v6 prompting becomes easier to compare when each submitted version has its own file.

The field should contain one coherent arrangement, without contradictory performance instructions.

5. Step 4 — Paste lyrics with clear section tags

Paste the following into Lyrics. Keep each bracketed heading on its own line.

[Verse]
The last train shakes the window frame.
Your empty cup still holds its place.
I turn the kitchen light down low.
And watch the headlights come and go.

[Pre-Chorus]
The clock keeps moving through the rain.
I stop rehearsing what to say.

[Chorus]
Leave one light in the night window.
Let it shine where the lost roads go.
I can find my way back slowly.
Leave one light, one light for me.

[Verse]
The morning waits behind the glass.
I let another minute pass.
Your folded coat stays by the door.
I do not count the steps anymore.

[Pre-Chorus]
The clock keeps moving through the rain.
I stop rehearsing what to say.

[Chorus]
Leave one light in the night window.
Let it shine where the lost roads go.
I can find my way back slowly.
Leave one light, one light for me.

[Bridge]
If silence has a place to end.
Then let this room begin again.

[Chorus]
Leave one light in the night window.
Let it shine where the lost roads go.
I can find my way back slowly.
Leave one light, one light for me.

[Outro]
One light for me.

Enter Night Window — A in Title. Save the same lyric block as lyrics-v1.txt.

Keep production prose out of the sung lines. Bracketed section labels are prompts, not guaranteed arrangement commands.

Do not add elaborate invented tags during this baseline. A short label makes failures easier to diagnose.

Your lyrics should show distinct sections, repeated chorus wording, and no unbracketed production instructions.

6. Step 5 — Generate and score the baseline

Check the model and controls again, then click Create once. Label the returned candidates A1, A2, and onward in your notes.

A request may return multiple candidates. Record the actual count rather than assuming every interface behaves identically.

Listen through each candidate once without editing anything. Then replay the first verse, first chorus, bridge, and ending.

Score each item using this scale.

• 0 points: The requested behavior is absent.
• 1 point: The behavior appears partially or inconsistently.
• 2 points: The behavior is clearly audible.

| Check | Maximum score | | --- | --- | | Sparse verse arrangement | 2 | | Fuller chorus arrangement | 2 | | Clear, centered lead vocal | 2 | | Drums removed during bridge | 2 | | Ending decays without truncation | 2 |

Record timestamps beside failures, such as 01:42 — bridge retains drums. Your Suno v6 prompting log should separate observations from guesses.

Do not score louder playback as better production. Keep the volume control fixed throughout this comparison.

A useful baseline should reveal specific successes and failures, even when no candidate satisfies everything.

7. Step 6 — Test section-level production notes

Duplicate lyrics-v1.txt and name the copy lyrics-v2.txt. Change only the section headings shown below.

| Original heading | Replacement heading | | --- | --- | | Each [Verse] | [Verse: sparse drums, muted synth pulse] | | Each [Chorus] | [Chorus: fuller drums, wide synth chords, backing harmonies] | | [Bridge] | [Bridge: no drums, sustained pad] |

Leave the actual lyric lines unchanged. Keep the same style text and control values.

Set Title to Night Window — B, then click Create once. Label the returned candidates B1, B2, and onward.

This tests a practical convention, not a documented parameter language. Suno may interpret, ignore, or occasionally vocalize bracketed instructions.

Score the new candidates with the same five checks. Compare the verse-to-chorus transition before comparing decorative details.

This Suno v6 prompting test changes one instruction layer. It does not hold the generated melody or performance constant.

Keep the expanded headings only if their results serve your brief. Restore the plain headings if instructions become audible lyrics.

A matching B candidate should make the requested section contrast easier to hear.

8. Step 7 — Fix one recurring production failure

Identify the largest repeated failure across A and B. Select exactly one correction from this table.

| Observed failure | Single change to test | | --- | --- | | Lead becomes heavily ornamented | Add Keep the lead melody syllabic, with minimal vocal runs. to Styles. | | Chorus remains thin | Replace the chorus sentence with Add sustained synth chords and backing harmonies in choruses. | | Unwanted guitar dominates | Enter distorted electric guitar in Exclude / Exclude Styles. | | Spoken passages interrupt singing | Enter spoken word, rap in Exclude / Exclude Styles. | | Ending cuts off sharply | Replace the ending sentence with Resolve the final phrase, then let the last chord decay. |

Do not apply several rows together. Save the changed style or exclusion text in test-c.txt.

Enter Night Window — C as the title and generate once. Preserve your chosen lyric version and all control values.

Exclusions remain requests, not hard filters. Keep them short and tied to an audible problem.

If no unwanted style appears, leave exclusions empty. Broad exclusions can remove characteristics you wanted to retain.

Your C candidates should provide evidence about one correction, rather than several simultaneous edits.

9. Step 8 — Run a controlled slider comparison

Choose the best A, B, or C configuration. Copy its complete input text into slider-baseline.txt.

For Suno v6 prompting, change sliders separately from text. Leave Variety at zero throughout these comparisons.

Generate the following three batches, without changing lyrics, style text, or exclusions.

| Batch title | Weirdness | Style Influence | Variety | | --- | --- | --- | --- | | Night Window — D | 20% | 60% | 0 | | Night Window — E | 20% | 90% | 0 | | Night Window — F | 35% | 75% | 0 |

D and E compare different Style Influence values against your original 75% baseline. F compares a higher Weirdness value against the original 20% setting.

These are exploratory comparisons, not controlled measurements of model behavior. Every generation can introduce unrelated musical differences.

Score all returned candidates before selecting a winner. Record the model and slider values beside each title.

If credits are limited, run D first and stop after scoring. Mark E and F as untested rather than inventing conclusions.

A completed comparison should show which actual candidate best matches your production brief.

10. Step 9 — Save a reusable prompt package

Select the highest-scoring candidate that also satisfies your three priorities. Break ties by choosing the clearer vocal performance.

Open its song page and copy its address into selected-take.txt. Record the candidate label, score, and unresolved defects.

Save these files together.

• final-style.txt, containing the selected style text.
• final-lyrics.txt, containing the selected lyric version.
• final-settings.txt, containing model, sliders, and exclusions.
• selected-take.txt, containing the song address and listening notes.
• session-notes.txt, containing the comparison history.

Add this line to final-settings.txt.

Reusing these inputs does not guarantee the same melody, voice, or arrangement.

Finish your Suno v6 prompting template with three replacement slots: genre, vocal character, and section contrast. Keep the tested settings beside them.

For another song, replace those slots and the lyrics. Start a new comparison log rather than treating previous scores as transferable.

Your folder should now contain a selected draft and the exact inputs needed for another test.

11. Troubleshooting

• Step 1: Core v6 is missing. Check the account plan and model dropdown. v6-mini is a different variant. If you continue there, label the entire session accordingly.
• Step 3: The submitted style text changes. Confirm Variety is set to 0. Restore style-v1.txt before generating another baseline.
• Step 6: Production notes become sung words. Restore plain section headings. Keep arrangement directions in Styles, then regenerate with unchanged sliders.
• Step 7: The exclusion removes useful instruments. Delete the broad exclusion and name only the unwanted element. For example, replace guitar with distorted electric guitar.
• Step 8: Higher Style Influence produces a less suitable take. Keep the better candidate. Generate another pair at the same settings before drawing conclusions about that slider.

12. Tools

Valkat Studio covers lyrics, Suno prompts, vocal profiles, cover art, and music videos. Its audio-finishing tools include the mastering suite.

13. Final thoughts

Keep the selected draft, exact inputs, and timestamped failures together. Your Suno v6 prompting template is a tested starting point, not a deterministic preset. Change one instruction layer per comparison, then judge the resulting audio.

14. References

• How to Use: Creative Sliders - Suno Help.
• suno.skin › en › featuresSuno Custom Lyrics — Write or Co-Write Your Words.
• Android Create: Custom Mode.
• iOS Create: Custom Mode - Suno Help.
• Can I use my own lyrics?.