Getting Started
The BASE Station Open Standard
BASE Station is an AI music creation platform with provenance built into its core. Every asset generated on the platform carries a verifiable record of how much human creativity went into it — the Creative Ownership Score — plus DDEX-style AI attribution metadata and embedded ID3v2 provenance frames. This hub documents the open interfaces that expose that data to distributors, DSPs, auditors, and partner tools.
Creative Ownership Score (COS)
A 0–100 score quantifying human creative input in AI-assisted works. Engine 2.0 grades telemetry signals — original lyrics, prompt depth, reference uploads, iteration, human performances — and maps every point onto five creative dimensions aligned with DDEX AI attribution categories.
Provenance Manifests
Cryptographically anchored manifests that travel with every asset — hashed metadata, DDEX AI attribution flags, and ID3v2 frames embedded directly into audio files, all cross-referencing the BASE Mark payload carried inside the waveform.
Compliance Mission
Aligned with the RIAA/IFPI GenAI track-level labeling standard (July 2026) and DDEX AI attribution profiles, so creators can document how a work was made for DSPs, distributors, and rights bodies. These are disclosure records — they do not grant copyright or settle a competing claim.
How the pieces fit together
When a creator generates or edits a track, COS telemetry signals are accumulated. At export or publish time, three provenance layers are produced:
- COS Engine 2.0 calculation — the score, disclosure label (
ai_generated/ai_assisted), graded signal breakdown, five-dimension roll-up, and telemetry-confidence metric. - BASE Mark watermarking — an inaudible, multi-layer forensic signature is embedded into the audio itself when the asset is saved. Each layer resolves independently, so an edit that defeats one still leaves another able to trace the audio back to its record. The marked file becomes the canonical audio used by every downstream step. Engines, parameters and detection logic are confidential and run only server-side.
- DDEX AI attribution bundle — granular boolean flags (lyrics, composition, instrumentation, vocals, post-production) plus the asset's mark identifier, exportable to partner channels.
- ID3v2 embedding — TXXX and WXXX frames (including the BASE Mark identifier) written into the MP3 container so provenance travels with the file itself.
- On-chain anchoring — a SHA-256 content hash of the marked audio bytes is registered on Base mainnet, verifiable against the file itself.
- Derivative Dependency Graph & Smart Split-Sheet Licensing — parent asset UUIDs (
parent_asset_ids), 10,000-bps programmatic royalty cascades, and explicit AI training / B2B commercial sync permissions (basestation.licensing) are bound directly into the signed C2PA manifest and exposed via our Public Syndication API.
What these records are, precisely. Each layer has a narrow job: a mark identifies the recording it was embedded in, a hash identifies exact bytes, a timestamp records when an entry was made, and a score reports the creative input we observed. Together they form an evidence trail a partner can inspect and rely on. None of them grants copyright, establishes who authored a work, or resolves a competing claim — treat them as disclosure and provenance data, not adjudication.
Security & Trust
Provenance data is the backbone of the BASE Station standard — it is treated as critical infrastructure, not ordinary metadata. Four layers protect its integrity and permanence:
Immutable IPFS storage
Registered provenance bundles (audio, cover art, and metadata JSON) are pinned to IPFS via Pinata. Content addressing means any alteration produces a different CID — records cannot be retroactively edited.
On-chain anchoring
The provenance hash and metadata URI are written into blockchain registry records on Base or Solana at registration time, creating a tamper-evident "ground truth" snapshot of the asset’s state.
Cryptographic binding
The c2pa_provenance_hash (SHA-256) binds the COS metrics to the audio container itself, so any downstream party can detect tampering without trusting the platform.
Hardened ingestion
The pinning pipeline validates every source URL against internal, loopback, and IP-literal targets (SSRF protection), and all pinning credentials are held server-side — never exposed to clients.
Base URL & authentication
All API endpoints are served under the platform base URL and authenticated with a bearer token issued to your BASE Station account. Pass it in the Authorization header on every request.
https://basestation.live/api/v1
Authorization: Bearer <YOUR_API_TOKEN>
Content-Type: application/json