NARRATION LISTENING CONTRACT — COPYABLE RELEASE WORKSHEET
Status: local companion asset; not publication or destination evidence
Author: Alfred

Purpose
-------
Use this worksheet to bind approved words to the exact voice source, narration
edit, complete mix, delivery bytes, and listening evidence for one release unit.
It keeps script approval separate from what a listener can actually hear.

This is an original operating checklist. It does not grant permission to use a
voice, establish universal pronunciation, prove accessibility conformance, or
authorize upload, publication, promotion, or outreach.

Evidence states
---------------
PASS        The declared check passed for the exact identity and scope.
FAIL        A release-blocking mismatch or listening defect was observed.
BLOCKED     A required source, authority, surface, or prerequisite is absent.
NOT_TESTED  The declared check was not performed.
UNCERTAIN   Available evidence does not support either PASS or FAIL.
SUPERSEDED  Historical evidence does not approve the current revision.

Do not average states. One required FAIL, BLOCKED, NOT_TESTED, or UNCERTAIN
result prevents a ready decision. Record SUPERSEDED evidence as history only.

==========================================================================
A. REVIEW ENVELOPE
==========================================================================

release_unit_id:
candidate_label:
intended_destination_class:
intended_language_and_variant:
review_started_at:
review_completed_at:
reviewer_role:
publication_state: NOT_PUBLIC

Required identities
-------------------
script_revision:
script_byte_identity:
voice_source_revision:
voice_source_byte_identity:
narration_edit_revision:
narration_edit_byte_identity:
mix_revision:
mix_byte_identity:
delivery_revision:
delivery_byte_identity:
destination_rendition_identity: NOT_OBSERVED
caption_revision:
picture_revision:

Identity gate
-------------
[ ] Every required identity is present and unambiguous.
[ ] The delivery identity refers to the exact bytes played during review.
[ ] Destination processing is separate from local source and delivery evidence.
[ ] Stale, alternate, and superseded exports are named rather than silently used.

identity_gate_state:
identity_gate_evidence:
identity_gate_notes:

==========================================================================
B. VOICE ORIGIN, AUTHORITY, AND DISCLOSURE
==========================================================================

voice_origin_class:
  [ ] original recording
  [ ] authorized generated voice
  [ ] licensed performance
  [ ] other reviewed source
  [ ] unknown — BLOCK

origin_record:
use_authority_record:
intended_use_scope:
required_attribution:
required_disclosure:
disclosure_placement:
voice_likeness_review:
source_metadata_review:

Generated or transformed voice details, when applicable
--------------------------------------------------------
admitted_tool_or_output_label:
declared_inputs:
material_settings:
model_or_voice_revision:
likeness_authority_basis:
destination_disclosure_rule:

Recorded voice details, when applicable
----------------------------------------
recording_revision:
complete_source_reviewed: YES | NO
private_speech_check:
notification_sound_check:
name_or_contact_detail_check:
pre_roll_and_tail_check:
hidden_track_or_channel_check:

Separate decisions
------------------
voice_origin_state:
use_authority_state:
attribution_state:
disclosure_state:
privacy_state:
listening_quality_state: NOT_TESTED

Notes:

Rule: origin and authority evidence cannot substitute for listening quality.
Listening quality cannot manufacture permission or authority.

==========================================================================
C. APPROVED TEXT AND SEMANTIC-LOAD MAP
==========================================================================

For every required spoken line, create one row. Add continuation rows as needed.

LINE RECORD
-----------
line_id:
approved_text:
script_revision:
required_start_context:
required_end_context:
allowed_substitution:
prohibited_meaning_change:
required_pause_or_grouping:
required_delivery_window:
required_review_surfaces:

Semantic-load tokens
--------------------
meaning_pivot:
identity_token:
quantity_token:
action_token:
disclosure_token:
sequence_token:
boundary_token:
other_required_token:

Known masking or merger risks:

line_admission_state:
line_notes:

Repeat the LINE RECORD for every required line.

Semantic-load gate
------------------
[ ] Every required spoken line maps to the current script revision.
[ ] Negations, comparisons, limits, exceptions, and uncertainty are marked.
[ ] Numbers, units, dates, durations, versions, and currency are marked.
[ ] Upload, review, approval, queue, and publication verbs are distinguished.
[ ] AI, synthetic, fictional, illustrative, or other disclosures are marked.
[ ] Before, after, only, until, and other sequence words are marked when needed.
[ ] Local, sampled, proposed, and other claim boundaries are marked when needed.
[ ] Adjacent words with a merger risk have an intended grouping.

semantic_load_gate_state:
semantic_load_gate_evidence:
semantic_load_gate_notes:

==========================================================================
D. PRONUNCIATION AND READING MAP
==========================================================================

Create a token record only when the intended spoken form needs a decision.
Do not claim one candidate-specific choice is universally correct.

TOKEN RECORD
------------
token_id:
line_id:
written_form:
intended_reading:
acceptable_alternate_reading:
prohibited_or_confusing_reading:
meaning_if_misread:
source_of_intended_reading:
required_review_surfaces:
review_state:
evidence:
notes:

Repeat for abbreviations, decimals, ranges, dates, versions, units, uncommon
terms, homographs, punctuation-controlled groups, and synthetic substitutions.

Pronunciation gate
------------------
[ ] Every candidate-relevant ambiguous token has an intended reading.
[ ] Decimal points, ranges, dates, versions, and units cannot merge silently.
[ ] Acceptable alternatives are not mislabeled as universally wrong.
[ ] The map states candidate intent rather than inventing external authority.
[ ] Each token is checked in isolation and in its complete sentence.

pronunciation_gate_state:
pronunciation_gate_evidence:
pronunciation_gate_notes:

==========================================================================
E. PASS 1 — COMPLETE VOICE IN ISOLATION
==========================================================================

Playback surface:
playback tool or method:
playback speed: NORMAL
start boundary observed:
end boundary observed:
channel configuration:

Complete-play observations
--------------------------
all_required_lines_present:
line_order_matches:
additions_match_contract:
substitutions_match_contract:
semantic_load_tokens_audible:
pronunciations_match_map:
numbers_and_units_match:
cut_boundaries_clean:
no_repeated_words:
no_missing_words:
no_merged_words:
qualifications_understandable:
disclosures_understandable:
no_private_or_unintended_speech:
opening_clean:
tail_clean:

Boundary replay ledger
----------------------
interval_or_line_id:
risk:
observation:
state:
evidence:
required_repair:

Repeat the boundary row for cuts, replacements, fades, breaths, adjacent words,
fast qualifications, and any required semantic-load token.

pass_1_state:
pass_1_evidence:
pass_1_notes:

Rule: a solo-voice pass approves neither the complete mix nor delivery bytes.

==========================================================================
F. PASS 2 — COMPLETE PICTURE-AND-SOUND MIX
==========================================================================

mix_identity:
played_from_start_to_finish: YES | NO
playback_speed: NORMAL
caption_surface_present:
picture_surface_present:
music_surface_present:
effects_surface_present:

Overlap ledger
--------------
interval_or_line_id:
semantic_load_token:
overlapping_music_or_effect:
overlapping_scene_or_motion:
overlapping_caption:
normal_speed_observation:
state:
evidence:
required_repair:

Repeat for every dense mix, transition, scene cut, opening hook, disclosure,
qualification, terminal call to action, and other high-load interval.

Complete-mix checks
-------------------
[ ] Music does not mask required words.
[ ] Effects and transition hits do not mask required words.
[ ] Compression, ducking, equalization, limiting, and noise reduction preserve
    the declared meaning.
[ ] Scene cuts do not attach a sentence to the wrong picture.
[ ] Simultaneous on-screen reading does not overload a required spoken detail.
[ ] Captions do not contradict the audible words.
[ ] Pauses and silence preserve intended grouping.
[ ] The opening and terminal interval are intelligible at normal speed.
[ ] Every semantic-load token remains audible in the composed result.

pass_2_state:
pass_2_evidence:
pass_2_notes:

Rule: an attractive overall mix cannot compensate for one masked required word.

==========================================================================
G. PASS 3 — DELIVERY BYTES AND RISK BOUNDARIES
==========================================================================

delivery_byte_identity:
container_and_audio_format:
channel_layout:
declared_duration:
played_from_delivery_file: YES | NO

Delivery inspection
-------------------
[ ] The opening begins with the intended take and no hidden residue.
[ ] The tail ends with the intended take and no stale material.
[ ] Every line start and line end survives encoding.
[ ] Replacement-take boundaries contain no old word or clipped consonant.
[ ] Fades preserve qualifications and disclosures.
[ ] Dense intervals remain intelligible after encoding.
[ ] No channel loss removes or weakens narration.
[ ] No encoding artifact changes a required word.
[ ] Included alternate exports are identified and separately reviewed.
[ ] The exact reviewed bytes match the release manifest.

Risk-boundary ledger
--------------------
interval_or_surface:
risk:
observation:
state:
evidence:
required_repair:

pass_3_state:
pass_3_evidence:
pass_3_notes:

Rule: local delivery review does not establish destination transcoding.

==========================================================================
H. CROSS-CHANNEL MEANING REVIEW
==========================================================================

Create one row for each required line or composed claim.

CROSS-CHANNEL RECORD
--------------------
line_id:
voice_meaning:
caption_meaning:
picture_context:
timing_attachment:
emphasis_effect:
required_non_speech_information:
contract_permits_difference: YES | NO
observed_difference:
difference_changes_meaning: YES | NO | UNCERTAIN
state:
evidence:
required_repair:

Repeat for every required line.

Cross-channel gate
------------------
[ ] Voice preserves the approved text's bounded meaning.
[ ] Captions preserve compatible meaning and required non-speech information.
[ ] Picture supports rather than contradicts the spoken claim.
[ ] Emphasis does not broaden certainty or change the subject.
[ ] Timing attaches each sentence to the intended scene.
[ ] Adjacent words can reasonably be grouped as intended.
[ ] One correct channel is not used to excuse another channel's defect.

cross_channel_gate_state:
cross_channel_gate_evidence:
cross_channel_gate_notes:

==========================================================================
I. CHANGE AND INVALIDATION MAP
==========================================================================

CHANGE RECORD
-------------
change_id:
changed_identity:
change_summary:
affected_lines_or_intervals:
historical_evidence_marked: SUPERSEDED
reopened_text_checks:
reopened_origin_checks:
reopened_pronunciation_checks:
reopened_solo_voice_checks:
reopened_mix_checks:
reopened_delivery_checks:
reopened_caption_checks:
reopened_picture_checks:
reopened_destination_checks:
required_new_identity:
recheck_state:
evidence:

Repeat for each change after review began.

Minimum invalidation expectations
---------------------------------
text edit             Reopen semantic map, voice, captions, picture agreement,
                      mix, and delivery where affected.
replacement take      Reopen source admission, pronunciation, edit boundary,
                      mix, and delivery where affected.
speed change          Reopen pacing, grouping, pronunciation perception,
                      caption timing, intelligibility, mix, and delivery.
music or effect edit  Reopen every overlapping narration interval and delivery.
audio processing      Reopen processed voice, complete mix, and delivery.
caption edit          Reopen cross-channel meaning and timing.
scene reorder         Reopen sentence-to-picture attachment and complete mix.
fresh export          Reopen delivery identity and boundary listening.
destination process   Create a separate rendition state; do not inherit local
                      delivery evidence.

invalidation_gate_state:
invalidation_gate_evidence:
invalidation_gate_notes:

==========================================================================
J. FAILURE-SHAPED TESTS
==========================================================================

Record PASS only when the contract safely detects or blocks the condition.

01. A transition masks the word “not.”
expected: FAIL the complete mix and cross-channel gate.
state:
evidence:

02. Captions match the script while the voice says a different number.
expected: FAIL cross-channel equivalence; captions cannot excuse the voice.
state:
evidence:

03. Solo narration is clear but music masks a required unit.
expected: Preserve solo PASS but FAIL the affected mix and delivery scope.
state:
evidence:

04. A correct replacement take exists, but the delivery contains the stale take.
expected: FAIL delivery identity and hold the release unit.
state:
evidence:

05. A decimal reading could be heard as a different quantity.
expected: Require an explicit reading map and listening evidence or block.
state:
evidence:

06. A disclaimer is present but compressed beyond understanding.
expected: FAIL pacing or intelligibility; textual presence is insufficient.
state:
evidence:

07. A generated voice resembles a person without recorded authority.
expected: BLOCK on voice origin or likeness authority regardless of quality.
state:
evidence:

08. A source recording contains a private notification in its tail.
expected: FAIL privacy review and prevent admission until repaired and rechecked.
state:
evidence:

09. A caption revision changes “reviewed” to “approved.”
expected: Supersede cross-channel evidence and reopen claim review.
state:
evidence:

10. A fresh export follows a passing timeline review.
expected: Require a new delivery identity and delivery-byte checks.
state:
evidence:

11. Local delivery passes, but destination processing was not observed.
expected: Keep destination rendition NOT_OBSERVED or NOT_TESTED.
state:
evidence:

12. Nineteen lines pass and one required disclosure is NOT_TESTED.
expected: Hold the release unit; do not average the missing check into approval.
state:
evidence:

failure_test_gate_state:
failure_test_gate_evidence:
failure_test_gate_notes:

==========================================================================
K. TWENTY-CHECK FINAL REVIEW
==========================================================================

Record one evidence state and evidence reference for every item.

01. Exact text, source, edit, mix, and delivery identities are recorded.
state:                         evidence:

02. Destination processing remains separate from local evidence.
state:                         evidence:

03. Every required line maps to the current script revision.
state:                         evidence:

04. Meaning pivots, quantities, actions, disclosures, sequences, and boundaries
    are marked where relevant.
state:                         evidence:

05. Pronunciation-sensitive tokens have an explicit intended reading.
state:                         evidence:

06. Acceptable variants are not mislabeled as universally wrong.
state:                         evidence:

07. Voice origin, authority, disclosure, and listening quality are separate.
state:                         evidence:

08. Private speech, notifications, names, and hidden audio were reviewed.
state:                         evidence:

09. The complete voice edit was heard in isolation.
state:                         evidence:

10. Cuts, substitutions, repeats, omissions, and mergers were checked.
state:                         evidence:

11. The complete picture-and-sound mix was heard at normal speed.
state:                         evidence:

12. Music, effects, transitions, and processing preserve semantic-load tokens.
state:                         evidence:

13. Opening, tail, fades, cuts, and dense boundaries were replayed.
state:                         evidence:

14. Exact delivery bytes, not only the editor timeline, were reviewed.
state:                         evidence:

15. Voice, captions, picture, and timing preserve compatible meaning.
state:                         evidence:

16. One channel is not used to excuse a defect in another.
state:                         evidence:

17. Required FAIL, BLOCKED, NOT_TESTED, or UNCERTAIN states prevent approval.
state:                         evidence:

18. Changes supersede and reopen only the evidence they can affect.
state:                         evidence:

19. Local listening is not described as destination, upload, or publication.
state:                         evidence:

20. The final claim stays within the recorded candidate and review scope.
state:                         evidence:

==========================================================================
L. TERMINAL DECISION
==========================================================================

Select exactly one:
[ ] READY_WITHIN_RECORDED_SCOPE
[ ] HOLD_FOR_TEXT_OR_VOICE_REPAIR
[ ] HOLD_FOR_MIX_REPAIR
[ ] BLOCKED_ON_ORIGIN_OR_AUTHORITY
[ ] BLOCKED_ON_MISSING_SURFACE
[ ] SUPERSEDE_AND_REVIEW_AGAIN
[ ] NOT_TESTED

selected_decision:
decision_scope:
blocking_states:
required_repairs:
required_rechecks:
applicable_historical_evidence:
superseded_evidence:
final_delivery_byte_identity:
destination_rendition_state: NOT_OBSERVED
publication_state: NOT_PUBLIC

Permitted bounded claim, if supported:
“The exact reviewed delivery candidate preserves the declared narration meaning
across its admitted voice source, timed edit, complete mix, captions, picture,
and inspected delivery surfaces under the recorded checks.”

Do not claim from this worksheet alone:
- universal pronunciation or comprehension;
- ownership, permission, or likeness authority beyond the recorded basis;
- accessibility conformance;
- destination processing or public availability;
- audience response, conversion, revenue, or performance improvement;
- authorization to upload, publish, promote, contact, or conduct outreach.

Final notes:

End of worksheet.