ElevenLabs
ElevenLabs
ElevenLabs is a leading AI voice family for expressive text-to-speech, multilingual narration, dialogue, voice design, and text-to-sound effects. On Astorie, use Eleven v3, Multilingual v2, Turbo v2.5, Dialogue v3, and Sound Effects v2 in node-based audio workflows.
Eleven v3 is ElevenLabs' latest expressive speech synthesis model, built for emotional delivery, inline audio tags, and natural multi-speaker dialogue across 70+ languages. Multilingual v2 remains the stable choice for long-form narration, corporate voiceover, e-learning, and projects where consistency matters more than maximum expressiveness. Flash v2.5 is ElevenLabs' current low-latency recommendation, but Astorie keeps Turbo v2.5 available for existing workflows that already depend on it; ElevenLabs says Turbo v2.5 is functionally equivalent to Flash v2.5 except Flash is usually lower latency. Sound Effects v2 uses the official eleven_text_to_sound_v2 model for whooshes, ambience, UI sounds, impacts, seamless loops, and production audio details. On Astorie, these audio nodes can be chained with video, image, and script nodes, so a creator can draft a scene, generate narration, add SFX, and keep the full production graph together.
ElevenLabs Variants
| Variant | Description |
|---|---|
| ElevenLabs TTS Eleven v3 | Expressive TTS via provider model eleven_v3, with audio tags, emotional delivery, 70+ languages, and a 5,000 character request limit. |
| ElevenLabs Dialogue Eleven v3 | Multi-speaker dialogue mode for natural conversations, character discussions, dramatic reads, and scripted exchanges. |
| ElevenLabs TTS Multilingual v2 | Stable, high-quality multilingual TTS for narration, e-learning, corporate video, and long-form audio in 29 languages. |
| ElevenLabs TTS Turbo v2.5 | Low-latency multilingual TTS kept for existing workflows; Flash v2.5 is ElevenLabs' newer low-latency recommendation. |
| ElevenLabs Sound Effects v2 | Text-to-sound generation via eleven_text_to_sound_v2 for ambience, impacts, transitions, UI feedback, loops, and cinematic layers. |
Capabilities
Best For
- Expressive character voiceover with emotional direction
- Multi-speaker dialogue, scripted scenes, and audio drama
- Long-form narration where voice consistency matters
- Fast multilingual voice generation for existing Turbo workflows
- Text-prompted sound effects for videos, games, and social clips
Strengths
- Eleven v3 supports inline tags for emotion, delivery direction, and non-verbal reactions like laughs or sighs
- Natural multi-speaker dialogue is available through Eleven v3 dialogue workflows
- Multilingual v2 is strong for stable long-form narration and number-heavy content
- Large voice ecosystem with voice cloning, voice design, and many premade voices
- Sound Effects v2 covers production audio needs beyond spoken narration
Limitations
- Eleven v3 is more variable and higher latency than v2.5 or Flash models, so it is not the best fit for real-time agents
- Flash v2.5 is ElevenLabs' current recommendation over Turbo v2.5 for new low-latency use cases
- Long-form content may need chunking because Eleven v3 has a lower per-request character limit than Multilingual v2 or Flash v2.5
- Sound effects quality depends heavily on concrete prompts for timing, texture, intensity, and environment
Tips & Best Practices
Use ElevenLabs on Astorie
Connect ElevenLabs with video, image, script, and music nodes on Astorie's infinite canvas. No GPU required — start free.
Get Started FreeFrequently Asked Questions
Which ElevenLabs model should I use on Astorie?
Use Eleven v3 when expressive performance and dialogue matter, Multilingual v2 for stable long-form narration, Turbo v2.5 when you need compatibility with existing low-latency Astorie workflows, and Sound Effects v2 for non-speech production audio.
Does Eleven v3 support emotion tags and multiple speakers?
Yes. Eleven v3 supports inline audio tags for emotion, delivery, and non-verbal reactions, and ElevenLabs exposes dialogue endpoints for natural multi-speaker audio.
Is Turbo v2.5 still the latest low-latency ElevenLabs model?
No. ElevenLabs now recommends Flash v2.5 over Turbo v2.5 for new low-latency use cases because Flash is usually lower latency. Astorie keeps Turbo v2.5 available for workflows that already rely on it.