ElevenLabs' Eleven v3 introduces Audio Tags, a feature in their new Text to Speech model that enhances character performance by allowing users to control tone, emotion, and pacing in speech. This tool enables precise direction over vocal identity, making it possible to switch accents, dialects, and archetypes like villains or narrators within a script without changing the underlying text or voice. This flexibility is ideal for applications such as animation, games, and interactive fiction, where character voice is crucial. Audio Tags allow for dynamic vocal changes and contextual shifts, transforming text into a performance that matches the desired persona. While Professional Voice Clones are not yet fully optimized in v3, users are encouraged to use Instant Voice Clones or designed voices during this research preview.