Last week, ElevenLabs unveiled the language expansion for its cutting-edge artificial intelligence (AI) text-to-speech (TTS) model. This enhancement allows the AI to support an additional 41 languages, bringing the total to 70 languages. The New York-based startup indicated that this development makes the model accessible to approximately 90 percent of the global population. The Eleven V3 (alpha) model was initially launched on June 8, marketed as the company’s most expressive TTS offering to date.
Eleven V3 Now Supports 70 Languages
In a recent update shared on X (formerly Twitter), ElevenLabs confirmed that their latest AI model, Eleven V3, has incorporated support for 41 new languages. This update enables the model to natively generate audio from text in a total of 70 languages. The newly introduced languages include Arabic, Assamese, Bengali, Bulgarian, Catalan, Gujarati, Latvian, Malay, Malayalam, Marathi, Nepali, Swahili, Tamil, and Telugu.
The company has recommended that users looking to create text in any of the new languages should record an Instant Voice Clone (IVC) while selecting their desired language. Furthermore, ElevenLabs is set to introduce Voice Library voices for these new languages in the upcoming weeks.
Eleven V3 succeeds the multilingual V2 and V2.5 TTS models. The latest version is designed to support inline audio tags, including whispers, excitement, sighs, and more. By incorporating these audio tags, the model is capable of delivering a more expressive audio output that captures emotional nuances and non-verbal cues, enriching the overall dramatic effect of the spoken text.
The model is also designed to facilitate multi-speaker interactions, accommodating interruptions, natural pacing, and overlapping dialogues. Additionally, ElevenLabs asserts that the V3 model has enhanced capabilities in managing factors such as stress, cadence, and contextual comprehension. Interested users can access Eleven V3 through the company’s website and mobile applications, although it is currently not available via an application programming interface (API).
In April, ElevenLabs launched a new enterprise-oriented feature known as Agent Transfer. This feature is a component of the company’s Conversational AI toolkit and allows two AI agents to engage in dialogue and exchange information. It establishes a framework where one AI agent can transfer a conversation to another, more specialized agent, carrying along all relevant conversation data.
