How to caption a video with your exact words and the audio's timings
Give Vizard Agent the script or lyric file and say the captions must use that wording exactly. It builds its own word-level transcript purely to find where each phrase is sung or spoken, then lays your text onto those spans, including in passages the recogniser could not hear, where your lines are placed at the actual vocal entrances.
What do you type into Vizard Agent?
Separate the two sources explicitly. Name where the text comes from and where the timing comes from as two different things, because by default both come from the same place, the recogniser, and that is exactly the arrangement you're trying to break. Vizard Agent aligns your text phrase by phrase, so send it in performance order: a lyric sheet missing a repeated chorus throws the alignment off from that point on.
Prompt
What does Vizard Agent actually do?
The recogniser is still used throughout, but only ever as a clock, never as a source of words. Its transcript says where in the audio each syllable falls, which is the one thing your script can't tell anyone. The words themselves always come from your file. In the session behind this article it:
- Inspected the existing captions on a music video for drift against the vocal, and built an independent word-timed transcript to compare against.
- Compared the opening captions with the first vocal entrances, then checked the later sections for drift that had accumulated.
- Read the word-timing structure to find the vocal spans, aligned the supplied lyric sheet onto them and rendered the result into the existing caption design.
- Isolated the quiet outro, the kind of passage where lines simply disappear, and analysed its sung phrase timing directly.
- Placed the supplied lines at the real entrances the recogniser had missed, then built the combined track and re-checked it end to end.
When does this not work well?
Alignment needs the text and the performance to agree. An improvised take, a dropped verse or a different edit of the song puts your file and the audio out of step, and from that point the alignment is guessing. Heavy overlapping vocals and instrumental passages give it nothing to attach a line to, and a translation needs the spoken language to align against.
Common questions
Why transcribe at all if I have the words? For the timings. Vizard Agent uses the transcript as a clock, not as text.
What format should my file be? A script, an SRT or a plain lyric sheet. Vizard Agent reads all of them.
Why did only the end drift? Usually a quiet passage. Tell Vizard Agent and that section gets its own analysis.