Watch a one-word correction in an AI caption

Pause at 1.60 seconds. The small caption over the waveform now reads “Um,” at the end of the first group. The large lettering above it belongs to the source recording.

Spoken transcript

“I like the blue label. Um, use the green one for print.”

The voice is synthetic. Both editing examples use the same four-second source recording.

What changed in the second pass?

The automatic caption read “and.” Replacing that word changed where the captions broke into phrases. Both four-second previews have 100 frames and identical decoded audio samples. To see exactly which words moved, compare the two previews.

The corrected caption is also visible in the finished MP4. Its final five seconds show the Valmera ending.

To see how a picture change can follow a spoken word, watch the cutaway on “green”.

Download this previewDownload the originalAI caption editing guide