Eleven v3 is a third-party audio generation model from ElevenLabs, available on Runway. It reads emotional nuance and pacing directly from your script, supports inline audio tags like [whispers], [laughs], and [shouts], and can generate speech in 70+ languages.
This article covers how to access Eleven v3 in Custom and Workflows, script and audio tag best practices, available settings, and what to expect from your generations.
Spec Information
| Output | Speech audio (.mp3) |
| Supported inputs | Text |
| Modalities | Text to Speech |
| Languages | 70+ supported languages Audio tags apply in English only |
| Voices | Same voice IDs as Eleven Multilingual v2, plus custom voices from your library Custom voices can be created in Custom ony |
| Maximum script length | 5000 characters |
| Credit cost | 1 credit per 50 characters of script |
Step 1 – Selecting Eleven v3
The model picker lives at the bottom of the generation panel. Eleven v3 is available in Custom and Workflows.
- From the left sidebar, click Custom.
- Select the Audio tab at the top of the generation panel.
- Click the model dropdown at the bottom of the panel and choose Eleven v3.
Step 2 – Writing the script
Eleven v3 performs your script rather than simply reading it. The model picks up emotion and pacing directly from your writing, so punctuation, phrasing, and formatting all shape the delivery. You can write in any of 70+ supported languages.
Using audio tags
Add inline tags like [whispers], [laughs], or [shouts] directly in your script to shape performance and emotion. Place each tag immediately before the text it should affect — the script box highlights recognized tags as you type.
Note: Audio tags apply in English only, and your script must include spoken words — a script containing only audio tags will not generate.
Choosing a voice
Choose a voice for your generation. Eleven v3 supports the same voice IDs as Eleven Multilingual v2, and you can also select any custom voice available in your library. To create a new custom voice, use Custom — the Workflows node supports selecting existing voices only.
Step 3 – Generating with Eleven v3
Before generating, review your settings:
- Script — the text you want performed, with any audio tags placed immediately before the words they should affect
- Voice — the preset or custom voice that will perform your script
Credit cost is driven by script length at 1 credit per 50 characters. See the Spec Information table for the full breakdown.
Once your script and voice are ready, select Generate.
Next steps
Once you've generated your audio with Eleven v3, put it to work:
- Act-Two — Drive a character's performance with your generated dialogue, matching lip sync and expression to the audio.
- Seedance 2.0 — Use your generation as an audio reference to pace and time a video generation to your voiceover.
- Agent — Hand your voiceover to Agent and describe the finished piece — it can generate the visuals, assemble the edit, and deliver a complete video around your audio.