Zen Zeppelin
Zen Zeppelin turns any song into a spoken-word meditation, narrated by AI and grounded in real research into what its lyrics actually mean.
Application
Sign back in — jump straight into a new meditation or manage the account
Environments and narrators
The Sanctum — a wood-paneled gondola drifting above the alps at dusk, aurora pressed to every porthole
Toolkit
We sing along to our favorite songs for years without understanding what they mean. I built Zen Zeppelin to ask that question — once you hear its answer, you won't hear that song the same way again.
The Context
Meditation apps reuse a fixed library. Zen Zeppelin researches and generates a fresh meditation for every song instead of templating one. Every meditation is genuine commentary: quotes stay under a hard fair-use ceiling, every meaning claim needs two independent sources, and every narration is watermarked before it reaches a listener.
The Approach
Four provider interfaces sit behind the pipeline — lyrics, meaning research, script, and voice — each one lives as swappable configuration. Every generated script runs through a hard validation gate: how much it quotes, how long any excerpt runs, how much commentary backs each line. Anything that fails triggers an automatic regeneration.
What I Built
A real, end-to-end pipeline: lyric retrieval, meaning research, script generation, voiced narration, and a deep-link hand-off into the listener's music service. Layered on top is what makes it feel like an experience: an auditionable cast of narrator voices, and seven original, hand-designed meditation spaces chosen before you press start.
The Outcome
A functional product, tested end-to-end on real songs, designed and built solo from the first line of code to the finished pipeline. The hard part — proving an arbitrary song reliably becomes a good meditation — is already answered. What's left is go-to-market infrastructure, pricing, and getting it in front of people.