BlackTape – 基于 MusicBrainz 和 Discogs 的开源音乐发现工具
我开发了一个音乐发现应用，现在是开源的。
BlackTape 使用 MusicBrainz 和 Discogs 这两个开放的、由社区维护的数据库，来索引艺术家并根据他们在各自流派中的独特性进行排名。艺术家越小众，他们的排名就越高。这与 Spotify 的算法正好相反。
我对推荐算法使发现变得单一感到沮丧。在超过 1000 万个索引艺术家中，总是出现相同的艺术家。MusicBrainz 记录了 260 万个艺术家，并提供丰富的流派标签、场景数据和地区元数据。Discogs 则有长达 80 年的发行元数据。将这两个数据库结合起来，根据独特性而非流行度对艺术家进行评分，发现的空间就完全打开了。
它的功能包括：
- 按流派/场景进行原子标签组合搜索
- 根据独特性评分排名的发现推荐（稀有 = 可被发现）
- 完整的艺术家页面：唱片目录、标签、相关艺术家、场景数据
- Spotify 播放集成（可选）
- 时间机器：按年代浏览艺术家
- 风格地图：可视化流派/场景导航
- 知识库：流派关系图
没有追踪，也不依赖平台 API 来获取核心发现数据。桌面应用使用 Tauri + SvelteKit 构建。
GitHub: https://github.com/AllTheMachines/BlackTape
网站: https://blacktape.org
欢迎讨论 MusicBrainz 流程、独特性评分或开放数据的发现方法。
查看原文
I built a music discovery app. It's open source now.<p><pre><code> BlackTape uses MusicBrainz and Discogs — open, community-maintained databases — to index artists and rank them by how unique they are within their genre. The more niche the artist, the higher they surface. It's the
inverse of how Spotify's algorithm works.
I got frustrated watching recommendation algorithms flatten discovery. The same artists keep surfacing out of 10+ million indexed. MusicBrainz has 2.6 million artists catalogued with rich genre tags, scene data, and
regional metadata. Discogs has release metadata going back 80 years. Combine those two databases and score artists by distinctiveness rather than popularity, and the discovery space opens up completely.
What it does:
- Search by genre/scene with atomic tag combinations
- Discover feed ranked by uniqueness score (rare = surfaceable)
- Full artist pages: discography, tags, related artists, scene data
- Spotify playback integration (optional)
- Time Machine: browse artists by decade
- Style Map: visual genre/scene navigation
- Knowledge Base: genre relationship graph
No tracking, no platform API dependency for the core discovery data. Desktop app built with Tauri + SvelteKit.
GitHub: https://github.com/AllTheMachines/BlackTape
Site: https://blacktape.org
Happy to talk about the MusicBrainz pipeline, the uniqueness scoring, or the open-data approach to discovery.</code></pre>