fb103ac5ae
Every registered string now lives in one table (codecs/registry.rs) with its kind, status, media class, registering SPEC section, edition and registration date. `wai registry export` writes registry/capabilities.json, its JSON Schema, the generated table in REGISTRY.md and vocab v2; --check fails on drift. vocab v1 stays stable. A candidate capability (a registered string whose payload format no revision has pinned) must not be emitted. The C packers, the browser writers and `wai wrap` refuse one, and `codecs::is_dispatchable` is the single dispatch and selection predicate (SPEC §4): candidates, companion-only capabilities, records and derivations are not dispatchable, and an unregistered string is decoded only where the sink advertises it. The codec glossary is linked to the registry through a reviewed table and held to its schema by a library test; `org` becomes `body`, from a closed allowlist. The SPEC gains §5 sections on kinds and candidates, a §8 edition rule, corrected citations and rewritten appendices. The quantum-ops vectors are regenerated with neutral device ids. CI declares the feature-gated test targets with required-features, checks the registry export, builds wai-web for wasm32, and runs two public-text checks (comparative prose, and private name patterns) with a self-test. The specification, extensions, research notes, READMEs, glossary and source comments are rewritten to name no company or product and to make no comparisons. Co-Authored-By: Claude Opus 5.5 <[email protected]>
4.7 KiB
MPEG-AI Part 6 — requirement mapping and alignment
A preparation document, not a submission and not a normative extension. It maps what WAI already specifies onto the requirements of an open Call for Proposals, so the tree can be aligned against that work rather than diverge from it by default. Submitting an input contribution requires SC 29 national-body membership; nothing here should be planned around an assumed submission.
1. The work item
ISO/IEC 23888-6 — Information technology — Artificial intelligence for multimedia — Part 6: Media authenticity and provenance indication. Carried as an ISO PWI project record. The Call for Proposals was approved by WG 2 at MPEG 155.
| milestone | when |
|---|---|
| CfP responses, as input contributions to MPEG 157 (Brisbane) | 13 January 2027 |
| CD / CDAM | MPEG 158 |
| DIS / DAM | MPEG 159 |
| FDIS / FDAM | early 2028 |
Scope, as published: system-level media authentication across the entire lifecycle — authoring, creation, production, editing, storage, delivery and playback — built on MPEG Systems technologies. Stage is exploration/requirements; there is no working draft, no bitstream, no format and no capability string to register. A contribution accepted in January 2027 has roughly a year of drafting influence before the text sets.
2. Why this is not already covered by WAI's provenance ingest
WAI's existing provenance work is authoring-side manifest provenance — a manifest asserting who made an asset and how, bound to it by hash. That is a different layer from a systems-layer delivery-and-playback authenticity specification, which is what this work item is scoped to.
The two are complementary, and treating the existing ingest as coverage would be a category error: an authoring manifest says what happened before distribution; this work item is about what can be asserted and checked during it.
3. What WAI has that the requirements space does not
Two positions in this tree are absent from every provenance standard surveyed in 2030-transport-landscape, and both sit squarely inside a lifecycle-scoped authenticity specification:
Per-object measured energy with a declared acquisition class. Provenance
standards record what was done and by whom. WAI's binding records what it
cost, as measured, with the provenance of the measurement stated: it is
per-object, carries HwShunt / ModelBased / Estimator, refuses to publish a
figure from a failed measurement, and binds the measured quantity to the media
object under signature.
Determinism tiers separating byte-exact reconstruction from approximate re-rendering. Authenticity over generated or reconstructed media needs a statement of what reproduction means for the object in question. A hard hash over rendered output is only meaningful when the reconstruction is pinned; WAI says when it is, and — since the tier fix — says that the tier covers what the sink emits rather than the decode alone.
These map onto lifecycle requirements at the delivery and playback stages, which are exactly the stages an authoring-side manifest does not reach.
4. Alignment actions, in order of cost
- Specify against the requirements, not around them. As WAI's provenance, energy and determinism extensions are revised, state how each satisfies or declines a published requirement. This costs nothing now and is the difference between a profile and an alternative later.
- Keep the vocabulary reconcilable. Where this tree names a concept the work item also names, prefer their term or state the mapping. Two vocabularies for one property is the failure mode already recorded for energy acquisition classes.
- Establish a membership path if a contribution is wanted. This is an organisational decision with a lead time, and the deadline is fixed.
5. The asymmetry that makes silence expensive
The downside of doing nothing is not neutral. If this lands as the systems-layer provenance format, WAI is either a profile of it or an alternative to it, and that is decided by whether the alignment work happened — not by whether a contribution was submitted.
Alignment is cheap and reversible. Being an alternative to a ratified systems-layer standard is neither.
6. What this document does not claim
- It does not assert that a contribution will be made, or that membership exists.
- It does not treat the CfP requirements as final; the stage is exploration/requirements and requirements move.
- It does not describe WAI's mechanisms as adopted by, aligned with, or endorsed by that work item. They are not. This is preparation for a conversation that has not happened.