Correct wording in both languages does not establish correct pairing in time. A Chinese cue and its English translation may have different segmentation, start times or durations. Pair them by the spoken meaning first, then adjust the display intervals. Moving an entire language track can hide one offset while leaving individual clauses mismatched.
Establish the source and translation relationship
CapCut’s bilingual subtitle guidance describes its recognition-and-translation workflow. The page refers to CapCut International. It is not evidence that Jianying or every account has the same feature. Regardless of the editor, review the approved source wording and translation separately from their timing.
Assign a pair identifier to each semantic unit. One short Chinese sentence may require a longer English phrase, while a long sentence can be split into several readable cues. Pair identifiers should follow meaning, not merely equal row numbers. A missing cue early in one track can otherwise shift every later row pairing.
Record source text, translated text, source interval and target interval in separate columns. Mark one-to-many pairs explicitly. Preserve intentional sound captions and speaker labels rather than forcing them into a dialogue row with no corresponding speech.
Test three pairs together
For a proposed sample, use three six-second instructional units. Prepare reviewed Chinese and English text, including one English translation that takes more space. Play both tracks together and compare each cue with its matching instruction. The aim is to test a pairing method, not to claim a translation service’s accuracy.
- Find the actual spoken start and end of each unit.
- Match every translated cue to that unit’s pair identifier.
- Adjust segmentation where the longer language needs more than one cue.
- Verify that the second cue still belongs to the same instruction.
- Check combined placement and reading time in the rendered sample.
Do not make two languages display for identical durations simply to satisfy a spreadsheet if that makes one unreadable. Preserve semantic alignment while choosing a supported presentation that gives the audience enough time. If the agreed layout cannot accommodate the approved translation, revise the layout or wording with approval rather than deleting meaningful content.
Accept the paired result
Acceptance means neither language presents the next instruction while the other still displays the previous one, and each approved statement corresponds to the real audio. Check the final pair as carefully as the first, because imported tracks can have unequal numbers of cues.
Keep the pairing table with the clean picture master and both text files. The plan is reproducible but has not been executed as a product benchmark. Imagild Skills can prepare illustrations for the sequence; translation tracks are arranged in an external editor. If the meaning itself differs, return to language review. Timing corrections should not disguise an inaccurate translation or rewrite approved dialogue without notice.