Eleven v4: directing delivery for voiceovers and dialogue

The same line can sound like an invitation, a warning or a weary observation. In a short film, that difference can define a character. A new voice model is therefore interesting through the direction of a line: not only what it reads, but why the character says it.

What ElevenLabs announced

Eleven v4 and its Turbo variant were introduced on September 28, 2026; the article was updated October 19. ElevenLabs describes delivery instructions, audio tags and support for more than 90 languages. These are provider claims. This story does not report our own speed or quality measurements.

Direction starts before generation

Imagine a character arriving late and saying, “Almost on time.” In our hypothetical scene, they could be apologetic, ironic or sincerely untroubled. The words stay the same while the intention changes. Write that intention down before choosing an impressive emotion.

It helps to describe the listener, what happened before the line and what response the character wants. These are scene questions rather than service settings. They give the creator a way to judge a take beyond whether the voice sounds expressive.

A short audition instead of a long take

Our proposed evaluation uses one short line in two versions. In the first, the character tries to sound confident; in the second, they hide embarrassment. Judge both against the same scene: which communicates the intention and fits the picture?

Avoid a pile of conflicting directions at the start. If the output misses the brief, clarify one point: pace, a pause or the attitude toward the listener. This is an editorial recommendation for trying alternatives, not a promise that every instruction will be followed precisely.

Listen for the useful details

• Is the character’s intention clear without an explanation?
• Do important words, names and stresses sound appropriate?
• Does the delivery fit the on-screen action?
• Does the chosen manner remain comfortable across several lines?

A creator’s takeaway

We find it useful to think of TTS as recording takes: establish the scene’s purpose, make alternatives and select. A calm narrator does not always need the most dramatic emotion. A considered pause and restraint may do more. The final question is whether the voice serves the story.