# PSALM > Agent-facing counterpart to the [human project page](/projects/psalm/). ## Record metadata - Record: 019 - Slug: psalm - Domain: Software - Domain code: SW - Type: Media app - Status: Dormant - Period: 2025–26 - Portfolio role: Product experiment - Publication state: Public case study planned - Case-study readiness: Needs current-state audit - Compendium edition: 0.4 ## Summary An Apple Music-oriented listening interface that filters unwanted album artwork while preserving the music workflow. ## Overview Music services combine audio catalogs with a visual environment the listener does not fully control. PSALM explored whether someone could keep the usefulness of a mainstream catalog while filtering cover art that violated a personal visual-content boundary. A music application and backend that identify unwanted cover art and replace or obscure it while maintaining an Apple Music-style browsing experience. Purpose: Make a mainstream music catalog more usable for people with personal visual-content boundaries. ## The problem behind the project Visual content can make otherwise acceptable music products unusable for some people. The experiment asked whether a preference layer could solve that narrow mismatch. The idea is narrower than building a new streaming service and broader than hiding a single image. It requires consistent catalog processing, clear behavior when classification is uncertain, and an interface that does not make the filter more distracting than the artwork it replaces. Listeners who want control over displayed artwork may benefit. Artists and catalog providers are affected by classification and presentation choices, so labeling and platform terms matter. ## How it took shape An iOS-style frontend, an Elitedesk-hosted backend, automated cover-art processing, and a classification pipeline that evaluated roughly 100,000 covers at low cost. The project paired an Apple Music-style iOS experience with an Elitedesk-hosted backend and an automated artwork-classification pipeline. Roughly 100,000 covers were processed at low cost, establishing technical plausibility without establishing a user market. Josiah identified the unmet need, designed the filter experience and system direction, and directed implementation and large-scale processing with AI assistance. The technical pipeline and interface worked, and the cover-processing experiment demonstrated plausible operating cost. The project did not secure meaningful testers. ## What the project means now PSALM's unresolved questions are mostly product and platform questions: whether the integration is permitted and maintainable, how mistakes should be reviewed, and whether enough listeners share the need. A successful pipeline is evidence of feasibility, not evidence of product fit. Classification errors, catalog rights, Apple platform integration, and weak user validation remain. It is not currently maintained as a live service. A technically affordable content layer still needs a reachable audience, clear platform permission, and transparent error handling. Audit the surviving code and decide whether to preserve it as a case study or revive it around a platform-supported integration. ## Publication and interpretation notes - Current classification: Dormant - Portfolio readiness: Needs current-state audit - Publication boundary: Public case study planned ## Additional agent context The scale figure concerns processed artwork, not active users. Do not imply Apple affiliation or current service availability. ## Related project records - [OpenFilter Lab](/projects/openfilter-lab/llm/) — A portable, local-first annotation format for user-controlled mute, skip, and visibility decisions in media. - [Subliminal Blocker](/projects/subliminal-blocker/llm/) — A browser-extension experiment for distinguishing subtle promotional influence from ordinary page content. ## Navigation - [Complete project index](/projects/llm/) - [Human version of this record](/projects/psalm/) - [About Josiah's working method](/about/llm/) - [Agent discovery map](/llms.txt)