OverNormal Play case study

My role

I designed and built the player, with help from AI: the listening pages, the playlists, the upload flow, the play counting and the admin tools. The songs in my screens are my own, published under my artist names.

The problem

Music made with AI tools needs somewhere to be heard. A catalogue like that is small, and the people making it want a clean place to share songs, show lyrics and see honest listening numbers.

The approach

The player is a web page, so there is nothing to install. A song page has the cover, the controls and the lyrics. Listeners can keep favourites in a playlist, open an artist page and share a link.

Plays are counted fairly. A play counts after 30 seconds of listening, the same song is not counted twice within five minutes for one listener, and requests from one address are rate limited. A page explains this to listeners.

Uploaders add their own MP3, with a cover, tags and lyrics, and tick a box to say they have the right to publish it. Reviews written by AI reviewer personas carry a badge that marks them as AI. Admin tools can hide or restore items, so a mistake can be undone.

Sign-in uses IDUnify, described in the IDUnify case study. For similar scope see the platforms and SDKs service.

Stack

PHP, MariaDB, Firebase and Apache, as listed on the demo page.

Screens

Public pages, my own songs

These are public pages captured signed out. Only my own song and my own artist name appear. Songs by other artists are not shown here.

What is public and what is not

Public: the feature list, the stack above and the public player pages. The songs are made with AI tools and published by the artist. Nothing is sold on the site.

Not public: listener data, admin screens and any figures about audience or revenue. I make no claims about either.

Related pages

• OverNormal Play on the demo page
• IDUnify case study
• Platforms and SDKs service
• All selected work