ElevenLabs Launches v4 and v4 Turbo Speech Models
New models feature improved expression control, lower latency, and support for 90 languages.
Quick Look
- ElevenLabs has released its v4 and v4 Turbo speech models, featuring enhanced expression control, faster voice cloning, and support for 90 languages.
- The update aims to improve conversational fluidity for enterprise voice agents and follows a period of rapid growth.
AI-generated summary
Why It Matters
ElevenLabs previously released its v3 model and raised $500 million from Sequoia earlier this year. The company has seen its annualized revenue run rate grow to over $600 million.
ElevenLabs launched two new speech models on Monday, called ElevenLabs v4 and v4 Turbo, offering more expression control, lower latency for voice agents, and support for more than 90 languages.
The company released its v3 model last year, and teased the model at an event in Warsaw earlier this year. For the v4 generation of models, ElevenLabs is adopting a new architecture that allows for better control and faster cloning. The company said that with v4, users will be able to clone a voice with just 10 seconds of audio.
On the creative side, the model handles voice identity better over longer chunks of text. It also keeps the context of the text in mind while reading it aloud to change expressions. ElevenLabs introduced inline tags to define expression with v3, and is expanding those tags in v4, letting users stack multiple tags and having the model follow the sequence.
The previous version supported 70 languages, and ElevenLabs has worked to get that number up to 90 languages with the new version. The startup said that it observed the biggest quality jump in Japanese, Brazilian Portuguese, Mandarin and Cantonese.
ElevenLabs has scaled its enterprise calling business rapidly over the last year, with more than 55% of its business coming from large companies. The company said that the new model is suited for voice agents, as the new version has lower latency to allow for more fluid conversation. Plus, the v4 can start generating audio as soon as the LLM behind it starts generating answers. What’s more, the model can also handle confrontations, escalations, and holds differently for better issue resolution.
Competition in speech models has ramped up as startups like Cartesia, Deepgram, Fish Audio, Boson, and WellSaid Labs have created expressive speech models. Big companies like Google and OpenAI have also improved their voice models.
ElevenLabs raised $500 million earlier this year from Sequoia earlier this year, in a round led by Sequoia that valued the company at $11 billion. There are already rumors of a followup fundraising round that would value the company at $22 billion. ElevenLabs annualized revenue run rate has climbed from roughly $330 million at the start of the year to over $600 million. The company has aggressively hired personnel in different markets such as India, Europe, and Brazil, as its headcount has reached over 800.
Open Questions
- Will the rumored $22 billion valuation round materialize?
- How will competitors respond to the new latency benchmarks?
