Bottom line:Claude benefits from long-context preference memory: profile-first, scenario-second delivers steadier playlist style.

Why this approach works

Most weak AI playlists come from under-specified prompts. When the prompt lacks constraints, outputs drift toward generic popular songs.

A stronger method is to define scenario, exclusions, version preference, and strict output format before generation.

Then run a pilot set first, validate quickly, and expand only after direction is correct.

Execution workflow

1. Define scenario variables (time, context, energy curve).
2. Set constraints (language ratio, era, exclusions, version preference).
3. Force output format: one line per song, "Song - Artist" only.
4. Test with a small pilot list, then scale and import to Apple Music.

Copy-ready Prompt Template

Use this as a base and replace scenario variables:

【我的长期偏好档案】
- 喜欢：华语流行、独立民谣、带城市氛围的英文流行
- 不喜欢：重金属、纯器乐、过度悲伤慢歌
- 版本偏好：录音室版优先
- 语言比例：中文60%，英文40%

【本次任务】
场景：周日夜间返程（90分钟）
情绪曲线：开场平静 -> 中段提神 -> 末段放松
输出数量：24首

【输出要求】
每行“歌名 - 歌手”，不要编号，不要解释，不要小标题。

Support Boundaries (important)

Playlist-link import supports Spotify, YouTube Music, Deezer, NetEase, QQ Music, KuGou, and Qishui/Douyin (douyin.com entry). Single-link conversion supports Spotify, YouTube Music, Deezer, and NetEase only; QQ Music, KuGou, and Apple Music links return unsupported.

Frequently Asked Questions

How many songs should I generate first?

Start with 15-20 as a pilot. Expand only after direction is validated.

Why enforce "Song - Artist" output?

It reduces ambiguity and improves import matching quality.

Should I optimize for one model only?

Not necessarily. Keep one master structure and tune model-specific phrasing.

Is small mismatch after import normal?

Yes. Cross-platform licensing/version differences are expected.

Continue Reading (AI Playlist Hub)

• AI Playlist to Apple Music (Main Guide)
• Text-to-Playlist Import Guide
• Playlist Sync Troubleshooting
• ClipTunes
• Song Link Converter