Archive — US — techcrunch
ElevenLabs’ new v4 speech model supports more expression control and 90languages
ElevenLabs launched v4 and v4Turbo speech models with enhanced expression control, lower latency for voice agents, and support for 90 languages. The new model allows voice cloning from just 10 seconds of audio, improves handling of longer texts, and maintains context. The company has seen rapid growth, with over 55% of business from large enterprises and aims for an IPO in the coming years.
Key points
- Launch of ElevenLabs v4 and v4Turbo models with better expression control.
- Support for over 90 languages, an increase from 70 in the previous version.
- New architecture allows for voice cloning from 10 seconds of audio.
- Improved handling of long texts and complex expressions using inline tags.
- Lower latency enhances fluid conversation for voice agents.
- Company's revenue rose from $330 million to over $600 million; strong enterprise focus.
- Increased competition from other startups and big tech companies.
- Potential follow-up funding could double the company's valuation.
Why it matters
- Growing demand for expressive speech synthesis in enterprise applications.
- The rise of AI-driven voice technologies aligns with 2025's IT Trends emphasizing natural interaction.
- Increased investment in voice tech reflects AI advancements and market potential.
- Competition among startups and tech giants highlights the rapid evolution of speech synthesis.
ElevenLabs' advancements in speech modeling position it well within the competitive AI landscape, responding rapidly to market demands while planning for expansive growth and future funding opportunities.