music-promptlisted

# Music Prompt Engineering Turn a vague musical request into a specific, repeatable prompt. Most AI music prompts fail the same way: they describe a *feeling* but not a *recording*. `a sad song about losing someone` names no instrument, no tempo, no voice, no production era — so the model fills those blanks with whatever is statistically average, and the result is competent and generic. Your job is to leave no blanks. --- ## The five slots Every style prompt you write fills all five. Coverage matters more than order. | Slot | Answers | Examples | |------|---------|----------| | 1. **Genre** | What kind of music? | `Pop`, `Lo-fi`, `Stadium rock`, `Bossa nova` | | 2. **Mood** | What should the listener feel? | `Melancholic`, `Triumphant`, `Restless` | | 3. **Vocal** | Who sings, or nobody? | `Female vocal`, `No vocals`, `Whisper` | | 4. **Tempo** | How fast, and how does it move? | `90 BPM`, `Half-time feel`, `Swung` | | 5. **Production** | What does the recording sound like? | `Tape saturation`, `Dry close-mic drums`, `Vinyl crackle` | **Slot 5 is the one that separates specific from generic, and it is the one most often skipped. Never skip it.** Assembled: ``` Lo-fi, nostalgic, no vocals, 78 BPM, dusty piano loop, vinyl crackle, soft tape saturation ``` --- ## Two fields, two jobs Nearly every text-to-music tool has two inputs. Route content correctly or both get weaker. | Field | Carries | Never put here | |-------|---------|----------------| | **Style / tags**