MyPart AI Agents Access
Developer access to MyPart's music intelligence through Replicate and AI Model Context Protocol (MCP):
The offerings are designed to help developers, platforms and creators find existing songs that fit visual content, natural-language descriptions and creative intent - without generating new music. Developers and creators can use its models to integrate image and video soundtracking, contextual recommendations, music discovery, and more into their products.
For Replicate:
MyPart currently exposes two main capabilities on Replicate:
-
Visual-to-Music Matching, which instantly suggests contextually relevant music to any image, video or collection of visual files, to elevate them into their most engaging form. It does so by using the visuals as input, with an optional textual hint to refine the results: https://replicate.com/myart-arch/visual-music-search
-
Prompt Song Search, which finds songs based on descriptive prompts covering mood, genre, instrumentation, lyrical themes, writing style, and various other creative criteria: https://replicate.com/myart-arch/prompt-music-search
For MCP:
-
Alongside the Replicate models, MyPart Labs released general MCP AI music agents that combine music understanding with conversational and automated creative processes. These agents are designed to interpret music, lyrics, visual rhythm, scene transitions, audience context and user preferences to support music search, matching and playlist generation: https://www.npmjs.com/package/songhunt
Both models deliver ranked music recommendations from MyPart’s partner catalogs, featuring real, human-made music, never AI generated tracks. This can serve as a building block for applications that need to understand what a scene or brief calls for musically, and play songs that create an emotional impact.