Video recap audio: keep the original sound or add narration?

Example

ORIGINAL EDITORIAL EXAMPLE — The Radio After the Move This example is a written plan only. All clips and timestamps below are invented planning examples, written as readable text.

Creator-supplied context: when he moved house, Lao Zhou could not bring himself to replace this old radio. Label it as creator context; no frame shows it. Illustrated observations: a switch produces no sound; the technician reseats a connector; radio noise then becomes audible. The footage shows no open case, no identified fault and nothing about whether the fix lasts.

Five-beat draft cut — source clock -> provisional output clock: 1. 00:12–00:18 -> 00:00–00:06 | radio on the bench, no dialogue Owner: narration. Job: context. Chinese narration: “搬家时，老周没舍得换掉这台收音机。” Reading gloss only: “When he moved, Lao Zhou could not bring himself to replace this radio.” Do not send the English gloss to the Chinese voiceover. 2. 00:42–00:47 -> 00:06–00:11 | Lao Zhou speaks Owner: original_dialogue. Job: none. Preserve the complete Chinese line: “早上还响着，搬过来就没声了。” (It worked this morning; after the move, nothing.) No narration over the line. 3. 01:10–01:14 -> 00:11–00:15 | switch click, then no radio response Owner: silence. Job: none. Preserve the opening click and then the wait; the pause is this beat’s main event, and the source audio stays in. Do not add “he turns the switch” narration or music to erase the wait. 4. 01:28–01:34 -> 00:15–00:21 | connector seated, first radio noise Owner: action_sound. Job: none. Let viewers hear the change. The moment shows one response; leave the repair story untold. 5. 01:52–01:58 -> 00:21–00:27 | Lao Zhou reacts Owner: original_dialogue. Job: none. Preserve his complete line: “还以为它也不肯跟我走了。” (I thought it did not want to move with me either.) Let the reaction finish; do not append an unsupported moral.

Avoid this invented explanation: “The technician immediately knew the internal circuit had failed. He opened the case and repaired it.” The footage shows neither the diagnosis nor the case being opened. A useful replacement is the one context line above, then room for the original sound.

Revision exercise — what does “There is too much narration” actually change? These are two alternative edits to the same five-beat plan; pick one, and everything stays on paper.

First distinguish unnecessary description from a false claim: “The switch clicks. There is no response. The technician reseats the connector, and radio noise appears.” Those lines describe the illustrated observations accurately. But laid over beats 3 and 4, they make the viewer listen past commentary to hear the click, the wait and the first noise. This edit promises an audible change, and repeating it in words covers the very sounds that show it. A spoken-description version made for a different audience need gets its own sound plan; this allocation serves this cut.

A. Keep the context job and remove unnecessary repeated naming: Original Chinese: “搬家时，老周没舍得换掉这台收音机。” Shorter Chinese candidate: “搬家时，老周没舍得换它。” Reading gloss: “When he moved, Lao Zhou could not bring himself to replace it.” The radio is clearly on screen, so “it” has a visible referent; the move and reluctance both remain. If the opening switches to another object, a close-up of a person or a standalone audio version, establish the referent again. The gloss is for reading only; the voiceover uses the Chinese line. Whether the TTS finishes safely before the original line is checked in the actual voiceover.

“Lao Zhou could not afford to replace it” would invent an economic reason. The word “either” in his closing remark likewise gives no grounds for adding that nobody keeps him company after the move, or that a relative died. Cut words, keep the facts.

B. If the author no longer needs that context stated explicitly, omit narration from the first beat: Keep the existing introduction of the radio and the five-beat order. Beat 2 still says “It worked this morning; after the move, nothing,” and beat 5 retains Lao Zhou’s full response. No commentator repeats the event for him, and viewers lose the explicit mention of his reluctance to replace the radio, which neither of his recorded lines states. This option keeps all the picture, so the illustrative edit stays at 27 seconds. The example leaves the first beat’s ambient sound unspecified: inspect the actual material before assigning its new sound owner, keep the source audio and add no invented room tone.

Both A and B keep the wait after the click in beat 3. That beat establishes the lack of response before the connector action and noise in beat 4. The draft cut omits the source interval from 1:14 to 1:28; the two shots play back to back, yet 14 seconds of source separate them, so neither narration nor captions should say one flick of the switch fixed the radio for good. Both options keep this draft’s edit points; in your own project, set each wait’s length from the actual pause in your material.

Name the selected option in the handoff. A replaces the corresponding narration text and pending narration captions, so the captions match the new short line. B removes that narration together with its captions. Keep the two original lines and their original-dialogue captions subject to actual audibility. This handoff records the decision; existing voice files and mixes are redone in a separately authorized stage.

What should this same edit put in captions? — a text illustration assuming A is selected Faithful captions make heard words readable; that repetition across sound and text is useful. Remove distracting extra commentary and keep the dialogue captions. These notes keep the five beats and assume A has been chosen.

1. Radio introduction: the narration caption uses the selected shorter Chinese sentence, “搬家时，老周没舍得换它。” Do not retain the old longer sentence or add a separate “An old possession he cannot let go of” emotional summary. 2. Lao Zhou explains: retain the original-dialogue caption “早上还响着，搬过来就没声了。” Do not paraphrase it as “Moving broke the radio”: the line gives the order of events and leaves the cause of the fault unstated. 3. Click and wait: nobody speaks here, so invent no dialogue caption. If a separate sound-description layer is chosen, “[开关声]” (switch click) is a candidate label. It must not replace the waiting interval or anticipate “It will be fixed in a moment.” 4. Connector and first noise: still no dialogue caption. An optional “[收音机杂音]” (radio noise) describes a sound; “[故障排除]” (fault resolved) would turn one response into a complete diagnosis and repair. 5. Lao Zhou reacts: retain “还以为它也不肯跟我走了。” Do not add a closing “Even old possessions feel the sorrow of leaving” card; let his personification finish in his own words.

The opening caption could use two display lines: 搬家时， 老周没舍得换它。 This is a reading-layout break within one continuous utterance, voiced as a single TTS block. Actual caption cues and their in/out points follow real speech and output time; generate them from the actual voiceover and timeline.

The bracketed sound labels are separate editorial candidates. Keep them apart from recorded dialogue: hand them to the separately chosen caption/manual-finishing workflow and keep them out of original_subtitles.json. Dialogue captions alone leave the wait and the audible change unexpressed; a silent-viewing version needs its own decisions about how to show those events.

If B is eventually chosen, remove the opening narration caption with the narration. The author could separately add a context card, which brings back the background B leaves unstated. That is a separate expression choice for the author to make deliberately.

Timing note: beat 1 contains six seconds of picture, and the narration has to end well inside it. The actual TTS must finish with a safe source-audio recovery window before beat 2. The row boundaries above are drafts; confirm real pauses and safe cut points from the material’s evidence.

Text-only handoff to an agent with Video Recap Skills configured: Use the authorized source and take all times from its actual evidence. Prepare the story/audio board and a provisional cut plan around the five units. Let the click lead into its pause. Keep the two full lines and the radio’s first sound, use only the confirmed moving-house context, and leave other units without narration. Stop at the editorial handoff: do not call TTS, render media or add BGM. List unresolved sound/timing evidence. After I choose the edit and the cut exists, use the validated output-time mapping for narration.

If selecting A, give a bounded revision request that touches only this line: This is a text-revision exercise for the radio plan above. Change only the first beat’s Chinese narration to “搬家时，老周没舍得换它。” Do not apply B. Preserve the viewer promise, five-beat order, picture ranges, two complete original lines, the wait after the click and the first radio noise. Add no poverty, loneliness or fault diagnosis. Return the candidate wording and affected expression/caption handoff notes in this reply only; do not write into existing artifacts. The first beat still has a context-narration job. Do not redesign the story because one sentence became shorter. Do not cut, run TTS, replace audio, mix or render; leave the real duration to be checked after voiceover. If the actual opening fails to identify the radio clearly, explain the pronoun ambiguity and keep the noun; clarity comes before brevity.