Porto voice-to-audio architecture (proposal)

Porto voice-to-audio architecture (proposal) An architecture diagram generated by Archify. Frontend app · Upload, player, voice and feed · Architecture component Frontend app Upload, player, voice and feed Porto API · Accounts, publishing and edits · Architecture component Porto API Accounts, publishing and edits Job queue · SQS, retries and cancellation · Architecture component Job queue SQS, retries and cancellation Audio worker · Ingest, render and validate · Architecture component Audio worker Ingest, render and validate ACE-Step 1.5 · Original audio + edit instruction · Architecture component ACE-Step 1.5 Original audio + edit instruction Speech + text planner · Transcribe; constrained edit plan · Architecture component Speech + text planner Transcribe; constrained edit plan PostgreSQL · Uploads, lineage, jobs; chain views · Architecture component PostgreSQL Uploads, lineage, jobs; chain views Private S3 · Uploads, stems and renditions · Architecture component Private S3 Uploads, stems and renditions CloudFront · Signed playback and byte ranges · Architecture component CloudFront Signed playback and byte ranges Aptos Move · Funded usage, allocation and USDC · Architecture component Aptos Move Funded usage, allocation and USDC HTTPS Enqueue Lease job Audio + plan Voice / text Metadata Submit / index Read / write Private origin Legend Frontend Backend Database Cloud Message bus External

Co-create

  • • ACE-Step selected; owner controls remix rights
  • • Tempo, pitch and supplied stems use precise tools

Playback

  • • Worker returns a file, not live model output
  • • App plays through a short-lived CloudFront grant

Settlement

  • • Move owns financial state; SQL holds projections
  • • Listening evidence stays private and off chain