Eleven v3 Prompts
ElevenLabs' expressive text-to-speech model — the script is only half the prompt; the performance direction is the other half.
About Eleven v3
Eleven v3 is ElevenLabs' deeply expressive text-to-speech model. It doesn't just read text aloud — it performs it, rendering emotion, hesitation, and emphasis with a range that puts it among the most expressive voice models available.
Its defining feature is inline direction. Audio tags like [whispers], [laughs], and [sighs] dropped into the script change the delivery at that exact moment, and it can voice multi-speaker dialogue where each character keeps a distinct emotional register. Punctuation is part of the performance too: ellipses become hesitation, dashes become interruption.
Which means prompting Eleven v3 is closer to writing stage directions than typing text. The same sentence, tagged and punctuated differently, comes out as a different performance entirely. The gallery below collects tested Eleven v3 prompts with their actual audio, so you can hear how each direction technique sounds before adopting it.
How to write Eleven v3 prompts
- 1Place emotion tags immediately before the words they should color — [whispers] Don't move. [nervously] I think it heard us. A tag drifts if it sits far from its target line.
- 2Punctuate for the ear, not the page: ellipses create hesitation, a dash cuts a thought short, short sentences read as urgency. Punctuation is your pacing control.
- 3Write speech the way people actually talk — contractions, false starts, filler words. Formal written prose read aloud sounds like a press release, not a person.
- 4For dialogue, label each speaker and give them different emotional registers — one calm, one agitated. Contrast is what makes a two-voice scene feel alive.
- 5Build an emotional arc instead of a mood jump: move from [curious] to [uneasy] to [terrified] across the script. A single hard cut between emotions sounds like a splice.
- 6Use capitalization for shouted emphasis sparingly — one WORD lands, a whole sentence in caps flattens back into noise.
Frequently asked questions
What do tags like [whispers] and [laughs] do in Eleven v3?
They are inline performance directions. Placed inside the script, they change how the surrounding words are delivered — a whisper, a laugh, a sigh, a nervous tremor — at that precise point in the audio. The tested prompts on this page show where placement matters.
Can Eleven v3 do multi-speaker dialogue?
Yes. It can render conversations with multiple distinct speakers, each holding their own tone and emotional register. Label the speakers clearly in your script and give each a different mood — the contrast is what sells the scene.
How do I control pacing and pauses in Eleven v3?
Mostly through punctuation and sentence rhythm. Ellipses produce hesitation, dashes produce interruptions, and short sentences quicken the pace. Combined with emotion tags, this gives you fine-grained control over the performance without any special markup language.
Where can I try these Eleven v3 prompts?
Eleven v3 is available on the ElevenLabs platform. Copy any prompt from this page, paste the script with its tags intact, and pick a voice — then adjust the tags to fit your own scene.