界面已本地化;技能介绍与说明文档按所选语言由 AI 自动翻译,代码、命令、路径和专业术语保留原文,以原作者文档为准。
探索技能
效率与协作 · SKILL.md

minimax-music-3

minimax-music-3

【效率与协作】Adapt score or song cues into MiniMax Music 3 prompts using vivid English musical sentences plus explicit genre, mood, vocals, instruments, BPM, lyrics, and structure fields.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

技能说明

GitHub

name: minimax-music-3 description: Adapt score or song cues into MiniMax Music 3 prompts using vivid English musical sentences plus explicit genre, mood, vocals, instruments, BPM, lyrics, and structure fields. disable-model-invocation: true author: Alan Guice (Badgids) license: Apache-2.0

MiniMax Music 3 Prompting

Target model: music-3.0.

Prompt grammar

Write the main prompt as vivid English sentences, not a comma-only tag list.

A useful order is:

  1. mood and genre or subgenre
  2. tempo when important
  3. vocal character, or state that it is instrumental
  4. narrative or scene function
  5. atmosphere
  6. two or three important instruments and production details

Use structured fields alongside the prose prompt when available: genre, mood, vocals, instruments, bpm, key, tempo, structure, and references.

For songs, keep lyrics in the lyrics field with clear section markers. For film score, use instrumental mode unless the cue explicitly calls for vocals.

Rules

  • Describe vocals as an audible character, not only "male" or "female".
  • Give the track a scene or dramatic function.
  • Keep the instrument list selective.
  • Use English for the main generation prompt unless the target workflow explicitly benefits from another language. Lyrics remain in their intended language.
  • Save under 04_generation/prompts/minimax-music-3/<cue-id>.md.

Done

The musical brief is readable by a musician and every structured field agrees with the prose prompt.

我的收藏