829a03b89f
The controller (oldbox) sent a data-model alignment proposal and asked for six
additive changes on this node, the validation protocol, and an honest report.
Six of the seven changes are done; the seventh is declared missing rather than faked.
Stage 0 - reachability verified exactly as documented: health 200, heartbeat 401
(the correct answer - auth is refusing), missing Host 400. Nothing else was sent.
source_unit_key (controller 5.1, the blocking item)
umbraaudio/source_units.py. sha256("uk1" | chapter_source_text_sha256 |
occurrence | text_sha256). Nothing positional in the key: "occurrence"
disambiguates repeated text in document order.
- 18,181/18,181 units keyed across all four projects, 0 collisions
- chapter 1 renumbered +100 -> 0 keys changed (the load-bearing proof)
- backfill 10,015 rows in 0.11s, second run stamps 0 (idempotent)
- alignment phase 3: 200/200 probe units resolved with used_index: false
- tolerates the older projects that have no chapters.chapter_index
Recorded at generation time in chapter_builder, BEFORE the rows are read and
BEFORE the marker is written, so every path that renders audio also produces
identity and the clip can name its source unit.
5.7 recorded: utterance_index is KEPT as a display ordinal and made
non-load-bearing. spec 21 §10.
Blob hashes and effect chain (5.2)
Markers now carry the full sha256 of the delivered bytes and the effects that
produced them. Verified the existing order is already correct - write_marker
runs AFTER apply_effects - and closed the remaining hole: a changed trait list
changed the hash with no recorded cause. sha256 verification is opt-in, since
hashing every file on every validation pass would re-read gigabytes.
Vocabularies published (5.4) - umbraaudio/definitions.py, 19 sets, 126 values,
each value carrying whether it is 'declared' (spec/controller) or 'extracted'
(read out of code), so a diff shows a difference of meaning, not of wording.
Alignment report tool (6.6) - scripts/alignment_report.py --phase all. Emits the
controller's envelope with the per-phase evidence shapes, and obeys the two
rules they set: no evidence means blocked, never pass; a field we cannot produce
is null PLUS a finding, never omitted. Phase 2 is deliberately 'fail' - the
three-way store disagreement is real, and a pass there would be a check that
passed by doing nothing.
Fingerprint discipline (5.5) - the drift detector reported its OWN fingerprint
change when source_unit_key was added rather than hiding it. Node profile
updated with the new fingerprints and an honest note that they moved.
Node report extended - per-project source-unit-key coverage, per-voice-asset
sha256 + object_key, definition_set_version, controller stage reached.
Six gaps found in the proposal (spec 21 §11), including the one they invited plus
five more. The sharpest: "hash at generation" ignores that audio is
post-processed afterwards, and our key is anchored to chapters.source_text - a
second copy of the manuscript - so any change to chapter storage silently
invalidates every key in that chapter.
5.6 deliberately NOT faked: personality has no versioned artifact here, only
unversioned LLM prose in speakers.json. Reported as phase 6b finding B-03
instead, because wrapping unversioned inputs in a version field would produce
artifacts claiming a lineage they do not have.
Answered the controller's two open questions (infra/controller-handoff-2026-10-03.md):
forward-only link for a revealed recurring character, and yes - a new edition may
contradict a frozen state, since this node's correction path IS a regeneration.
Full suite 484 passed / 8 failed - the same 8 pre-existing and unrelated failures.
ruff clean on every file touched. The parallel session's chapter_builder hunk
(helper_url) was filtered out of the commit and left intact in the working tree.
212 lines
7.4 KiB
Python
212 lines
7.4 KiB
Python
"""Published vocabulary sets for the Umbra system (controller §5.4).
|
|
|
|
The controller's rule, from ``manifest.md`` §14.2, is that anything a stored row depends on
|
|
must be a *versioned vocabulary in data*, not a literal in code. Otherwise every product
|
|
reimplements it and they drift silently. This module is the audio side of that.
|
|
|
|
Each set records, per value, where the value came from:
|
|
|
|
* ``declared`` - stated by a spec or by the controller. This node does not get to change it.
|
|
* ``extracted`` - read out of executable code in this repository, and therefore changeable
|
|
without a deployment.
|
|
|
|
That distinction is the point of publishing them: a diff against the controller shows not
|
|
only that two sides disagree, but whether a disagreement is one of wording or one of meaning.
|
|
|
|
Nothing here imports the database or the filesystem. It is pure data, so it is safe to read
|
|
from any product and cheap to diff.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
from typing import Any, Literal
|
|
|
|
Origin = Literal["declared", "extracted"]
|
|
|
|
DEFINITION_SET_VERSION = "defs-2026-10-03"
|
|
|
|
# Where each vocabulary's authority lives, for the reader of the diff.
|
|
AUTHORITY = {
|
|
"utterance_type": "specs/16 §5a, umbraaudio/database.py SCHEMA default 'narration'",
|
|
"character_persistence": "controller D11 (specs/21 §9)",
|
|
"character_kind": "specs/18 §5.4a",
|
|
"voice_capture_mode": "umbraaudio/generation/chapter_builder.py, controller §2.2",
|
|
"phonetic_provenance": "manifest.md §14.2",
|
|
"attribution_evidence_kind": "umbraaudio/analysis/clarity.py",
|
|
"artifact_kind": "manifest.md §11, specs/20 §F.19",
|
|
"artifact_status": "controller §2.1 (artifacts.status)",
|
|
"project_lifecycle": "manifest.md §1.4, controller §2.1 (projects.status)",
|
|
"migration_state": "manifest.md §14.1",
|
|
"job_capability": "specs/21 §4.7 + controller §2.3",
|
|
"attribute_absence_reason": "controller D13",
|
|
"voice_variant_reason": "controller D6",
|
|
"name_token_status": "specs/17 §3.4",
|
|
"name_token_kind": "specs/17 §3.4",
|
|
"post_tts_effect_trait": "umbraaudio/generation/audio_effects.py TRAIT_EFFECTS",
|
|
"emotion": "umbraaudio/generation/chapter.py EMOTION_OPTIONS",
|
|
"pace": "umbraaudio/generation/chapter.py PACE_OPTIONS",
|
|
"qa_issue_type": "umbraaudio/analysis/attribution_qa.py",
|
|
}
|
|
|
|
|
|
def _d(*values: str) -> dict[str, Origin]:
|
|
return dict.fromkeys(values, "declared")
|
|
|
|
|
|
def _e(*values: str) -> dict[str, Origin]:
|
|
return dict.fromkeys(values, "extracted")
|
|
|
|
|
|
VOCABULARIES: dict[str, dict[str, Origin]] = {
|
|
# --- unit and unit-adjacent ------------------------------------------------
|
|
"utterance_type": _d("narration", "dialogue", "scene_break"),
|
|
# controller D11: persistence is DECLARED, never inferred from name overlap
|
|
"character_persistence": _d("book_local", "series_persistent"),
|
|
# specs/18 §5.4a. Note "speaker" is deliberately absent: it is a derived fact
|
|
# (a character with lines in a work), never a kind. Controller §1.2.
|
|
"character_kind": _d(
|
|
"named_character",
|
|
"recurring_role",
|
|
"scene_role_instance",
|
|
"group",
|
|
"ai",
|
|
"narrator",
|
|
),
|
|
# How a voice's identity was captured, per manifest.md §5.4
|
|
"voice_capture_mode": _e("embedding", "icl"),
|
|
# manifest.md §14.2 - IPA derivation is Latin-only, so provenance is load-bearing
|
|
"phonetic_provenance": _d("explicit", "ipa_derived"),
|
|
"attribution_evidence_kind": _e("name", "role", "pronoun", "address_only", "none"),
|
|
# --- artifacts -------------------------------------------------------------
|
|
"artifact_kind": _d(
|
|
"source_document",
|
|
"character_profile",
|
|
"voice_reference",
|
|
"voice_prompt_vmatrix",
|
|
"utterance_wav",
|
|
"chapter_stitch",
|
|
"m4b",
|
|
"mp3_track",
|
|
"ar_package",
|
|
),
|
|
"artifact_status": _d("intermediate", "final", "superseded"),
|
|
# --- projects and migration ------------------------------------------------
|
|
"project_lifecycle": _d(
|
|
"active", "on_hold", "released", "archived", "experimental", "parsed", "analyzed"
|
|
),
|
|
"migration_state": _d("legacy", "mirroring", "dual_write", "authoritative", "retired"),
|
|
# --- work routing ----------------------------------------------------------
|
|
"job_capability": _d(
|
|
"audio.parse_manuscript",
|
|
"audio.segment_utterances",
|
|
"analysis.character_discovery",
|
|
"analysis.speaker_attribution",
|
|
"analysis.speaker_clarity",
|
|
"analysis.name_token_scan",
|
|
"character.personality_analysis",
|
|
"character.consistency_check",
|
|
"voice.capture_character",
|
|
"voice.reference_generation",
|
|
"voice.attribute_measurement",
|
|
"audio.render_utterances",
|
|
"audio.stitch_chapters",
|
|
"audio.normalize_loudness",
|
|
"export.m4b",
|
|
"export.mp3_package",
|
|
"review.speaker_QA",
|
|
),
|
|
# controller D13: absence is typed, never NULL
|
|
"attribute_absence_reason": _d(
|
|
"not_attempted", "not_applicable", "in_progress", "superseded"
|
|
),
|
|
# controller D6: a variant with no reason is indistinguishable from a bug
|
|
"voice_variant_reason": _d(
|
|
"sarcasm_mimic", "disguise_roleplay", "technical_degradation"
|
|
),
|
|
# --- review ----------------------------------------------------------------
|
|
"name_token_status": _e("unresolved", "assigned", "created", "dismissed"),
|
|
"name_token_kind": _e("name_tag", "speaker_name"),
|
|
"qa_issue_type": _e(
|
|
"speech_tag_mismatch",
|
|
"narrator_dialogue",
|
|
"role_tag_guessed",
|
|
"address_mismatch",
|
|
"narration_speaker_mismatch",
|
|
"long_monologue",
|
|
),
|
|
# --- delivery --------------------------------------------------------------
|
|
"emotion": _e(
|
|
"neutral",
|
|
"calm",
|
|
"serious",
|
|
"sad",
|
|
"happy",
|
|
"angry",
|
|
"fearful",
|
|
"excited",
|
|
"gentle",
|
|
"firm",
|
|
"hesitant",
|
|
"sarcastic",
|
|
"tender",
|
|
"urgent",
|
|
"dramatic",
|
|
"whisper",
|
|
"romantic",
|
|
"commanding",
|
|
"melancholy",
|
|
"hopeful",
|
|
"desperate",
|
|
"surprised",
|
|
"confused",
|
|
"disgusted",
|
|
"determined",
|
|
"tired",
|
|
"confident",
|
|
),
|
|
"pace": _e("slow", "moderate", "fast"),
|
|
"post_tts_effect_trait": _e(
|
|
"raspy",
|
|
"gravelly",
|
|
"hoarse",
|
|
"breathy",
|
|
"whispery",
|
|
"nasal",
|
|
"deep",
|
|
"boom",
|
|
"reverb",
|
|
"echo",
|
|
"radio_filter",
|
|
"distortion",
|
|
"pitch_shift",
|
|
"noise_inject",
|
|
"tremolo",
|
|
"chorusing",
|
|
),
|
|
}
|
|
|
|
|
|
def vocabularies() -> dict[str, list[str]]:
|
|
"""The plain ``{set_name: [values]}`` shape the controller diffs (phase 1)."""
|
|
return {name: sorted(values) for name, values in VOCABULARIES.items()}
|
|
|
|
|
|
def provenance() -> dict[str, dict[str, str]]:
|
|
"""Per-value origin plus the authority each vocabulary is accountable to."""
|
|
out: dict[str, dict[str, str]] = {}
|
|
for name, values in VOCABULARIES.items():
|
|
entry = {v: o for v, o in values.items()}
|
|
entry["_authority"] = AUTHORITY.get(name, "undeclared")
|
|
out[name] = entry
|
|
return out
|
|
|
|
|
|
def declaration() -> dict[str, Any]:
|
|
"""The full published definition set, versioned."""
|
|
return {
|
|
"definition_set_version": DEFINITION_SET_VERSION,
|
|
"product": "umbraaudio",
|
|
"vocabularies": vocabularies(),
|
|
"provenance": provenance(),
|
|
}
|