ElevenLabs has released the alpha version of Eleven v3, its text-to-speech (TTS) model. This new version introduces significant advancements in AI-driven voice generation.
A key feature of Eleven v3 is its support for the Japanese language, addressing a long-standing demand. Furthermore, it incorporates emotional and conversational capabilities, setting it apart from previous models.
Eleven v3 expands language support from 33 to over 70 languages, increasing global population coverage from 60% to 90%. The model also features dialogue mode for natural speaker transitions and voice tag support for nuanced expression.
Users can now add vocal cues like sighs and laughter. The model also supports diverse emotional expressions within a single text. Streaming support is planned for the near future, targeting call centers and real-time conversational agents.
A free trial is available, allowing users to test the capabilities of Eleven v3. The free plan allows the generation of up to 20,000 characters per month, with a maximum of 2,500 characters per generation.