Launching today
Expressive Mode for ElevenAgents
AI voice agents that adapt tone, timing & emotion by context
70 followers
AI voice agents that adapt tone, timing & emotion by context
70 followers
Expressive Mode is a voice agent so expressive that it blurs the line between AI and human conversation. Powered by Eleven v3 Conversational and a new turn-taking system for better-timed responses with fewer interruptions.
@ElevenLabs has launched Expressive Mode for its ElevenAgents!
TLDR
The upgrade lets support bots sound calm, firm, or empathetic on command while timing replies more like a human.
It runs on a new speech model and a smarter turn-taking system, scaling to 70-plus languages for global call centers.
What's new?
Expressive Mode gives companies fine-grained control over tone, so an agent can de-escalate an angry traveler or speed through a clear set of instructions.
The feature is powered by Eleven v3 Conversational, which tracks dialog context and injects emotion without sounding fake.
A parallel turn-taking engine built on Scribe v2 Realtime reads pacing, volume, and intonation to decide when the AI should speak or pause.
Together they produce fluid, emotionally aligned voices that work across dozens of languages and dialects, including nuanced Hindi and regional Spanish accents.
The system is production-ready inside ElevenAgents with monitoring, integrations, and testing for large-scale deployments.
Updates I found note-worthy:
Fine-tune tone on the fly to reassure, clarify, or direct customers.
Context-aware TTS keeps emotional coherence across multiple turns.
New turn-taking logic reduces awkward interruptions and long silences.
Supports 70 + languages, improving nuance in under-served dialects.
Built-in to the existing platform with enterprise-grade reliability and analytics.
With this update, @ElevenLabs has pushed the bar for human-like AI conversations. What do you think?
Watched the airline customer support demo and it genuinely stopped me. The moment the agent shifts tone to meet the emotional state of a frustrated traveler, without sounding scripted or patronizing, is the thing that separates this from every other voice AI I've heard.
The parallel turn-taking engine is the unsung hero here. Knowing when to pause, when to hold silence, and when to step in is what makes a conversation feel human. Most voice agents get this so wrong that it poisons the whole interaction regardless of how good the voice quality is.
Curious how much control builders have over the emotional range. Can you define hard limits on how far the agent goes toward, say, warmth or urgency? Or is it fully contextual and automatic? That dial would matter a lot for brands with strict tone guidelines.
Omg thats amazing! This is exactly what's missing from Voice AI models today. This is game changer and could be finally the thing thats needed to help with outbound calling agents for sales 🫣
Adjust Page Brightness - Smart Control
congrats guys for this launch