Glossary
Temporal annotation
Labels tied to time ranges in audio or video — when an action starts and ends, who is speaking, where a scene changes. It turns raw footage into structured training data.
Annotations attach labels to time ranges: action boundaries, speaker turns, scene changes, on-screen events. They are produced by human annotators with tool assistance and checked by agreement between independent annotators. Dense temporal labels multiply the training value of the same raw footage.
Why it matters
Labelling is the expensive step, so footage that arrives already time-annotated saves buyers real money.