# Distinguishing comic text in ManYue: speech, captions and sound effects

Language: en
Canonical: [Distinguishing comic text in ManYue: speech, captions and sound effects](<https://allluckly.cn/en/guides/manyue-speech-caption-sound-effects/>)
中文：[中文 HTML](<https://allluckly.cn/guides/manyue-speech-caption-sound-effects/>)
English: [English HTML](<https://allluckly.cn/en/guides/manyue-speech-caption-sound-effects/>)
Published: 2026-11-03T00:00:00+08:00
Modified: 2026-11-03T09:34:11+08:00
Author: Bison

Use balloon pointers, placement and context to distinguish speech, captions and sound effects, retaining uncertain attribution and checking the source before quotation.

English translation of the [Chinese original](<https://allluckly.cn/guides/manyue-speech-caption-sound-effects/>), translated on 2026-11-03. Original publication dates, attribution, code and version notes are retained.

Not every written mark in a comic represents spoken dialogue. Captions, inner speech and sound effects can have different forms or be deliberately mixed. Attribute the text in a short sequence for more accurate quotation and story evidence. ManYue provides viewing and return navigation; no image-text recognition, spoken performance or automatic translation is assumed.

## Choose a sequence with different text roles

Choose one or two authorised pages containing clear balloons, caption boxes and embedded effects where available. Read them intact and record language/edition, since translations may alter fonts or placement. Do not force a sign into the sound-effect category to complete a checklist. Inner monologue or annotation can remain another unresolved role. The categories help investigate this sequence rather than impose a universal font taxonomy on the entire work.

## Check speakers from balloon pointers

Inspect the pointer and neighbouring responses before assigning a speaker. Off-panel figures, absent pointers or crowded scenes may remain uncertain. Proximity and balloon colour alone are insufficient. Record character labels and panel locations separately from quoted wording rather than adding an unverified name inside the quotation. Revise later when clearer responses provide evidence, preserving the reason for correction instead of disguising an initial guess as established fact.

## Do not equate captions with the author

A box may establish time/place, recall a character’s experience or present unreliable narration. Inspect person, tone and image relationship, recording information before attribution. No balloon does not automatically mean the author speaks, and the voice can change within a passage. Keep identity unresolved when needed and inspect nearby self-identification or memory cues. Preserve these when quoting so a character’s misunderstanding does not become an objective plot fact.

## Record sound effects with their visual role

For large embedded lettering, inspect its relation to action, impact and movement before interpreting sound, rhythm or mood. Size and distortion can contribute meaning beyond the transcribed word. Mark unfamiliar language unverified rather than presenting guessed pronunciation as translation. Compare authorised editions separately without altering images. Describe placement and apparent effect as your observation, not an explanation produced by the app.

## Separate signs and annotations from narrative voices

Signs, phone screens, translator notes and marginal remarks may belong to neither speech nor effects. Check whether text is a depicted object or has an annotation marker and edition explanation. Do not assign supplementary knowledge automatically to a character or original author. Keep location and attribution cues. Zoom to inspect small text, but mark unreadable source detail rather than guessing after repeated processing. Unreliably read text should not support a decisive plot claim.

## Keep a return anchor for each observed role

Bookmark relevant pages and record panel, role, attribution confidence and minimal necessary wording externally or in supported remarks. Screenshots are not required for every item. Use landmarks for tall images and restore the full page on return. This manual classification is not an automatic app tag. Preserve the file and recheck references after edition changes.

## Test the consequences of uncertain attribution

Take the least certain line and consider another speaker or narrator. If the story conclusion changes substantially, attribution needs further checking before the conclusion is fixed. Look for independent action evidence rather than making the ambiguous line carry everything. Retain provisional alternatives until later identification. The aim is not universal doubt but awareness of source-sensitive claims; an explicit uncertainty is stronger than detached text stripped of its balloon context.

## Acceptance and quotation check

Finish with evidenced roles, explicit uncertain voices and recoverable locations. Balloon shapes do not have universal fixed meanings, and no automatic sound-effect translation is claimed. Before quoting, recheck panel and edition and separate source wording, visual description and interpretation. Leave unusual forms unresolved instead of forcing attribution to the author; later evidence can then update the record without recycling an early misreading.

## Related guides and sources

[Referencing a panel in ManYue: create a verifiable address within a page](<https://allluckly.cn/en/guides/manyue-panel-reference-address/>); [ManYue small lettering: separate zoom level, source detail and display processing](<https://allluckly.cn/en/guides/manyue-small-lettering-check/>).

Checked on November 3, 2026: [Apple public version information](<https://itunes.apple.com/lookup?id=6753895889&country=cn>) lists ManYue 2.0.4, requiring iOS / iPadOS 15.0 or later. Feature scope follows the [official App Store description](<https://apps.apple.com/cn/app/id6753895889>) and the installed version. This article describes actions and content checks you can perform; it does not claim step-by-step physical-device or private-server testing. Manual records, external tools and application controls are identified separately, and code observations or personal inferences are not presented as verified automatic capabilities.

