ElevenLabs Launches Expressive V4 Speech Model in 90 Languages
ElevenLabs has launched Eleven v4 and Eleven v4 Turbo, introducing expanded expression controls, improved voice cloning, and support for more than 90 languages. The release targets content creators, developers, and businesses building AI-powered voice applications.
The new models also focus on conversational AI. Meanwhile, ElevenLabs aims to improve how AI voices handle emotion, context, and natural dialogue.
Eleven v4 Adds Advanced Expression Controls
Eleven v4 uses a new architecture to improve speech quality and emotional delivery. Consequently, users can create voices that convey urgency, humour, sadness, or excitement.
The model also expands the inline expression tags introduced with v3. In particular, creators can combine multiple tags to control delivery, pacing, reactions, and vocal style.
Moreover, Eleven v4 maintains speaker identity across longer passages. It also considers surrounding text when generating speech, helping voices respond naturally to context.
ElevenLabs says its Instant Voice Cloning feature can now reproduce a voice using just 10 seconds of audio. However, the company also offers Professional Voice Cloning for higher-fidelity results.
V4 Turbo Targets Real-Time AI Conversations
Eleven v4 Turbo focuses on applications that require rapid responses. For example, developers can use it to build conversational AI agents and customer support systems.
According to ElevenLabs, Turbo delivers median inference latency of approximately 100 milliseconds. Furthermore, it can begin generating audio while the underlying language model produces its response.
The company also highlights improved handling of interruptions, escalations, and customer interactions. As a result, businesses can develop voice agents that respond more naturally during complex conversations.
Expanded Language Support and Availability
Both models support more than 90 languages, up from approximately 70 in the previous generation. Meanwhile, ElevenLabs reports notable quality improvements in Japanese, Brazilian Portuguese, Mandarin, and Cantonese.
The models also support multilingual voice generation while preserving speaker identity. This feature could help businesses localise content and customer interactions across different markets.
Eleven v4 and v4 Turbo are available through ElevenCreative, ElevenAgents, and ElevenAPI. Additionally, developers can access the models through the company’s speech-generation infrastructure.
The launch comes as ElevenLabs expands its enterprise business. According to the company, large businesses now account for more than 55% of its business. Therefore, the latest release places greater emphasis on enterprise voice applications and real-time conversational systems.
