ElevenLabs has launched Eleven v3 Alpha, its most expressive and multilingual text-to-speech model to date. This new version brings advanced emotional control, improved naturalness, and multi-speaker dialogue capabilities — redefining realism in AI voice generation.
🔊 Key Highlights:
🌍 Global Reach: Supports 70+ languages, broadening its use across industries and regions.
🎭 Inline Audio Tags: Add nuance with in-text annotations like
whispers,laughs, orexcited— e.g.,happily``shouts We did it! laughs.👥 Text to Dialogue API: Generate multi-speaker audio using structured JSON — ideal for podcasts, audiobooks, or role-based simulations.
🧠 Improved Realism: Enhances stress patterns, cadence, and emotional flow for more human-like delivery.
💸 Launch Promo: Now available in the ElevenLabs app with 80% usage discount until June end.
⚙️ Access: Public UI access is live; API access by request. Real-time usage still recommended via v2.5 Flash or Turbo for now.