Files

B.D. Murphy 9bda553311 Add a speaker-review audit: is the attribution data fit for translation?

Runs after a reviewer finishes a book and answers the two questions that
gate the next stage:

  1. Are there utterances with no speaker?
  2. Now that speakers have been defined, renamed, merged or deleted, are
     there utterances that should change?

Read-only - it never edits the database, and a test asserts the files are
byte-identical afterwards. Output is a markdown table and a CSV, one row per
finding with chapter, utterance number and a text preview, so each item is
locatable rather than a count.

Deal Hunter, after the completed review (56 speakers, 0 still needing
review, 4319 utterances):

  - Dialogue with no speaker: 0. The 1803 speakerless rows are narration
    and scene breaks, which is correct and is reported as a number rather
    than silently dropped.
  - Utterances attributed to a deleted speaker: 0.
  - One speaker stored under two spellings: 6 rows across 3 speakers
    (narrator/Narrator, Kainda/Princess Kainda, Willamina Wake/Captain).
    All resolve to one voice so the audio is correct, but a per-speaker
    export would split them into six - caught before the translation stage,
    not after.
  - 6 speech-tag mismatches where the narration immediately before the line
    names a different speaker.
  - 10 dialogue rows voiced by the narrator: 9 are quoted labels ("Reactor.",
    "operations") that are correctly read by the narrator, 1 (ch28 utt46) is
    a real segmentation error. Classified advisory so the one real defect is
    not buried under nine false positives.

build_alias_map moved from scripts/qa_attribution.py into attribution_qa.py
so the CLI and the audit cannot disagree about what a speech tag resolves to;
qa_attribution output verified byte-identical.

13 new tests. Suite: 721 passed, 8 pre-existing.

2026-10-11 21:57:58 -07:00

5.4 KiB

Raw Permalink Blame History

Speaker review audit - DealHunterShip_4.0e_whole-to-audio

Generated 2026-10-12 04:57:45Z. Read-only: nothing was modified.

Summary

Utterances: 4319 (1874 dialogue)
Speakers: 56 (116 distinct names + aliases)
Narration rows with no speaker: 1803 - expected, narration is not spoken by a character
Findings needing a decision: 12

Dialogue still voiced by the narrator (10)

Chapter	Utt	Severity	Speaker	Detail
10	21	advisory	Narrator	Dialogue line still attributed to the narrator.
12	120	advisory	Narrator	Dialogue line still attributed to the narrator.
15	122	advisory	Narrator	Dialogue line still attributed to the narrator.
15	124	advisory	Narrator	Dialogue line still attributed to the narrator.
15	126	advisory	Narrator	Dialogue line still attributed to the narrator.
19	116	advisory	narrator	Dialogue line still attributed to the narrator.
25	178	advisory	Narrator	Dialogue line still attributed to the narrator.
25	187	advisory	Narrator	Dialogue line still attributed to the narrator.
27	14	advisory	Narrator	Dialogue line still attributed to the narrator.
28	46	advisory	Narrator	Dialogue line still attributed to the narrator.

Narration names a different speaker (6)

Chapter	Utt	Severity	Speaker	Expected	Detail
9	21	action	Kainda Walters Foster Murphy	Duchess Victoria	Nearest tag: 'Vicky says assertively.'
20	229	action	Warbots	The Warbot	Nearest tag: 'The Warbot asks,'
24	112	action	Unnamed Officer	Someone	Nearest tag: 'Someone says,'
25	132	action	Warbots	The Warbot	Nearest tag: 'The Warbot replies in a simple, neutral voice,'
26	196	action	Distorted Voice	Ruler	Nearest tag: 'Ruler says,'
26	200	action	Distorted Voice	Ruler	Nearest tag: 'Ruler replies,'

One speaker, several spellings in the database (6)

Chapter	Utt	Severity	Speaker	Expected	Detail
-	-	action	Captain	Willamina Wake	3 utterances stored under this spelling; the same speaker also appears
-	-	action	Willamina Wake	Willamina Wake	21 utterances stored under this spelling; the same speaker also appear
-	-	action	Kainda Walters Foster Murphy	Kainda Walters Foster Murphy	793 utterances stored under this spelling; the same speaker also appea
-	-	action	Princess Kainda	Kainda Walters Foster Murphy	78 utterances stored under this spelling; the same speaker also appear
-	-	action	Narrator	Narrator	15 utterances stored under this spelling; the same speaker also appear
-	-	action	narrator	Narrator	614 utterances stored under this spelling; the same speaker also appea

Role tag guessed as a named character (3)

Chapter	Utt	Severity	Speaker	Expected	Detail
6	158	advisory	Kainda Walters Foster Murphy	The Voice	Nearest tag: 'The voice calls again,'
6	212	advisory	Nanny	The Robot	Nearest tag: 'The robot says,'
26	152	advisory	Warbots	The Speaker	Nearest tag: 'Every Warbots lower their weapons, and the speaker says,

Narration speaker disagrees (pronoun) (13)

Chapter	Utt	Severity	Speaker	Expected	Detail
10	116	advisory	Brutus	It	Narration names 'It' as speaker
11	20	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
17	77	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
20	123	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
20	131	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
20	160	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
20	174	advisory	Player Four	He	Narration names 'He' as speaker
25	153	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
30	84	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
30	150	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
30	162	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker
31	12	advisory	Interrogator	As Kainda	Narration names 'As Kainda' as speaker
32	149	advisory	Kainda Walters Foster Murphy	She	Narration names 'She' as speaker

Addressed to the speaker themselves (3)

Chapter	Utt	Severity	Speaker	Detail
11	166	advisory	Service Robot	Dialogue addresses 'Service Robot' but speaker is also 'Service Robot'
31	109	advisory	Willamina Wake	Dialogue addresses 'Willamina Wake' but speaker is also 'Willamina Wak
31	111	advisory	Willamina Wake	Dialogue addresses 'Willamina Wake' but speaker is also 'Willamina Wak

Provenance

Generated by scripts/speaker_review_audit.py, read-only. Re-run with:

python scripts/speaker_review_audit.py --project DealHunterShip_4.0e_whole-to-audio

The CSV alongside it in the project (not committed, project dirs are gitignored) carries one row per finding including the full text preview.

5.4 KiB Raw Permalink Blame History