Glossary
Word-level timestamps
Updated · By the WithTranscript team · How we checked
Word-level timestamps: Word-level timestamps give the start (and often end) time of every word in a transcript, rather than one time per caption line or paragraph.
Speech-to-text engines produce them as a by-product of recognition. Caption tracks usually have one time per cue of a few words to a few seconds; auto-generated YouTube captions also time individual words internally, which is why they appear word by word on screen.
They are what makes precise features possible: highlighting the exact word being spoken, cutting a clip that starts on a word, searching for a phrase and jumping to the second it is said.
For reading or pasting into an AI, a timestamp per paragraph is usually enough, and word timings would only add noise.
How WithTranscript handles it
WithTranscript uses fine-grained timings to highlight the sentence being spoken and to jump to any line you click, but shows and copies one timestamp per paragraph.