10. Future experiments
← Common pitfalls · Guide contents
The methods in this guide began with things I heard while making tracks. Some have repeated often enough to shape the way I work, but that does not make them controlled evidence about Suno’s internal operation.
One question keeps returning: does a clean, restrained first generation give a complex remix more to hold onto than asking for the full destination immediately? Foggy Hearts suggests that exposed musical relationships can survive a Cover well, but one successful source is not enough to turn that into a rule.
I would compare it like this:
- Direct: generate the complex destination from the first prompt.
- Scaffolded: generate a restrained source, then Cover it into the same destination.
- Keep lyrics, musical brief, model version and relevant settings as consistent as the workflow allows.
- Compare the results without revealing which route produced which track.
I would listen for vocal naturalness, arrangement clarity, melodic coherence, temporal stability, genre authenticity and audible artifacts. I would also repeat the comparison across several prompts and seeds. One striking pair can suggest the next experiment, but it cannot tell me how reliable the effect is.
Other questions I want to test include:
- testing which Bridge properties survive reliably across contrasting genres;
- comparing the same prompt through Cover and Generate;
- changing one influence control at a time;
- revisiting successful prompts when Suno changes model version;
- recording interesting failures alongside successful results.
I am not looking for a hidden formula. I want a creative practice that can explain why it made a choice, keep the failures that challenge it, and adapt when the tool changes.
Appendices
Section titled “Appendices”Appendix A — Prompt Lexicon
Section titled “Appendix A — Prompt Lexicon”The Prompt Lexicon collects the semantic fields I use, the jobs I normally give them and several prompt shapes I return to.
Appendix B — Descriptor Library
Section titled “Appendix B — Descriptor Library”The Descriptor Library is where I collect musical directions, combinations and tensions worth exploring. I treat it as a moodboard, not a dictionary of guaranteed model behaviour.
Appendix C — Example Prompts
Section titled “Appendix C — Example Prompts”Example Prompts preserves the standalone recipes that informed the early vocabulary and fusion work. Project-specific prompts stay with their song breakdowns because the decisions around them matter as much as the final text.