Eleven v3 Audio Tags, a feature of the new Eleven v3 (alpha) Text to Speech model, enable users to control AI speech by adjusting tone, emotion, and pacing to match real-world contexts, providing situational awareness to the AI. These tags, which appear as words in square brackets, serve as performance cues that allow the AI to adapt its delivery mid-sentence, transforming narration into a performance that reflects emotional beats or situational shifts. This innovation is particularly valuable in dynamic or high-context scenes, such as sports commentaries or suspenseful audiobooks, where tags like [EXCITED], [WHISPERING], or [SHOUTING] can dramatically influence how content is perceived. The model's ability to shift tone mid-line and handle interruptions without rewriting scripts offers a new layer of creativity for voice designers, game developers, and storytellers, though Professional Voice Clones are not yet fully optimized for this version.