Eleven v4 Replaces SSML with Contextual Audio Tags for AI Voiceover Pauses

Eleven v4 replaces the SSML syntax used in earlier models like Multilingual v2 with Audio Tags, allowing writers to insert stage directions directly into scripts. This shift moves away from mechanical, pre-defined silence gaps toward contextual pauses that adapt to the scene's emotional tone.

Three Methods for Controlling Pauses

The model offers three distinct ways to manage pacing and silence:

• [pause]: Creates a short beat, roughly equivalent to a natural breath or a brief separation between ideas. It is ideal for adding emphasis after a setup line or preventing a period from feeling too rushed.
• [long pause]: Generates deliberate silence for dramatic tension, scene breaks, or reveals. Unlike fixed timers, the duration fits the narrative context.
• Punctuation: Ellipses, dashes, and short sentences serve as pacing instructions. An ellipsis slows delivery to create hesitation, while fragmented sentences add snap and directness.

Consistency and Workflow

For long-form audio, maintaining consistent pause conventions across the entire script is critical to avoid jarring output changes. Users can manually apply these tags or use the v4 Enhance button, which automatically inserts Audio Tags into the text. The system supports over 17,500 voices from the Voice Library, enabling rich vocal performances through these refined timing controls.