AI Music Creation Platform Build

I am undertaking a full-scale build of an AI-driven music creation platform—think Suno, but entirely my own. The heart of the product is music creation, with text-to-music generation as the flagship capability, so every architectural and UX decision should keep that at the forefront.

Scope of the product
The platform must ship simultaneously on the web plus native Android and iOS, all powered by a single, scalable backend that hosts the AI models and audio pipelines. Users will be able to type or paste a prompt, receive complete songs with AI-generated lyrics, vocals and instrumentation, then refine the result through style, genre and mood controls. Continuation, remix, cover versions, stem separation and basic in-browser/mobile editing tools all need to feel seamless.

Core feature set (deliverables)
• Text-to-music & text-to-song generation engine
• Lyric generator with custom-lyrics override and inline editor
• Multi-voice vocal synthesis and instrumental tracks
• Style/genre/mood selectors, plus song continuation & extension logic
• Remix, cover and stem-separation workflows
• Audio upload for processing, remix or voice cloning
• Non-destructive music editor (trim, splice, arrangement)
• User accounts, personal music library & generation history
• Web app (responsive React or similar)
• Native Android & iOS apps sharing code where practical (e.g., React Native, Flutter)
• Tiered, autoscaling backend (model orchestration, queueing, storage, CDN) with API for all clients

Acceptance criteria
1. Songs generate reliably under real-world load with latency acceptable for end users (