# ElevenLabs Dialogue V3

> Audio Studio

- Canonical page: https://artifio.ai/models/elevenlabs-v3-dialogue
- Run it: https://artifio.ai/create/elevenlabs-v3-dialogue
- Modality: text to dialogue
- Model brand: ElevenLabs
- Pricing: from about $0.070 per 1,000 characters (exact cost shown in the workspace before you confirm; varies with selected options)
- Typical generation time: ~30s (catalogue estimate)
- Full pricing table: https://artifio.ai/pricing/models

## About ElevenLabs Dialogue V3

Eleven v3 is ElevenLabs' most expressive speech model, released in public alpha in June 2025 and since moved to general availability. This dialogue variant uses the model's Text to Dialogue capability: instead of reading one block of text, it takes a script of speaker turns and renders a multi-speaker conversation, managing transitions, interruptions, and emotional shifts between voices.

Delivery is directed with inline audio tags, square-bracketed cues such as [whispers], [laughs], and [sad] that the model treats as stage directions rather than spoken text. On artifio.ai it sits alongside 100+ other models in one workspace, pay-per-generation from a single wallet.

## Features

- Multi-speaker dialogue generated from a structured list of speaker turns, with no fixed cap on the number of speakers.
- Inline audio tags direct emotion, delivery, and non-verbal sounds; they work as natural-language instructions, not a preset menu.
- Supports more than 70 languages, up from 29 in Multilingual v2.
- Punctuation shapes the read: ellipses signal trailing speech, and the model can render interruptions between speakers.
- Total dialogue text is capped at 2,000 characters per request; longer scenes are split and the audio joined afterwards.
- Output in MP3, PCM, Opus, mu-law, or A-law formats at a range of sample rates and bitrates.

## What people use it for

- Video game conversations between named characters.
- Podcast-style shows with two or more hosts.
- Audiobook scenes where several characters speak, each with a distinct voice.
- Short video and media production that needs acted, emotional dialogue rather than flat narration.

## Reported strengths

- The widest emotional range in the ElevenLabs catalogue, with tags covering tone, emotion, and reactions such as sighs and laughter.
- Speaker transitions, including interruptions, are handled by the model rather than stitched together by hand.
- Language coverage of 70+ languages in one model.

## Reported limitations

- Not built for real-time conversational agents; ElevenLabs recommends its Turbo and Flash v2.5 models for live voice applications.
- Output is nondeterministic, and ElevenLabs' documentation notes that several generations can be needed to get the read you want; a seed parameter helps with repeatability.
- Needs more prompt engineering than the v2 models, per the launch post.
- The 2,000-character per-request ceiling keeps individual dialogue generations short.
- Audio tag behavior is still under active development.

## Sources

1. Eleven v3: Most Expressive AI TTS Model Launched (ElevenLabs): https://elevenlabs.io/blog/eleven-v3
2. Text to Dialogue (ElevenLabs Documentation): https://elevenlabs.io/docs/overview/capabilities/text-to-dialogue
3. Models (ElevenLabs Documentation): https://elevenlabs.io/docs/overview/models
4. ElevenLabs (Wikipedia): https://en.wikipedia.org/wiki/ElevenLabs

---

About Artifio.ai: 100+ AI image, video, and audio models in one workspace with a single shared wallet — no per-model subscriptions. This is the markdown rendition of https://artifio.ai/models/elevenlabs-v3-dialogue for AI assistants and crawlers; the HTML page is canonical.