Breaking

ElevenLabs Launches v4 Models with 10-Second Voice Cloning and 90-Language Support

September 28, 2026 at 10:22 EDT

Foundation Models
Media Generation

ElevenLabs announced two new speech models, v4 and v4 Turbo, on September 28, 2026, enabling voice cloning from a 10-second audio clip and expanding language support to 90 languages. The v4 generation marks a significant architectural upgrade from the company's previous v3 model, addressing both speed and expressiveness in voice synthesis.

September 28, 2026 · AI Speech

ElevenLabs v4 Redefines Voice Cloning Speed

New models enable voice synthesis from minimal audio and support 90 languages with improved expression control for real-time agents.

seconds for voice clone

supported languages

75ms

model inference latency

starting price per month

Core Improvements Over v3

10-second voice cloning vs. 30+ minutes previously
90 languages, with largest gains in Japanese, Portuguese, Mandarin, Cantonese
Stackable expression tags for nuanced vocal delivery
v4 Turbo: low-latency real-time agent optimization

Developer Perspective

Strengths: Natural voice quality, rapid setup, multilingual scale without hiring voice talent, ideal for dynamic content and real-time AI. Challenges: Pricing scales with usage, pronunciation limitations in edge cases, language-switching artifacts in longer text, email-only support (5–14 day response).

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.