ElevenLabs has raised $19 million and introduced new products as it builds out its AI voice platform. The company’s plans include tools for creating long-form audio and detecting audio generated by its own technology, amid questions about how synthetic voices can be used and who may be affected.
Funding will support research and new products
Entrepreneurs Nat Friedman and Daniel Gross co-led the Series A round with Andreessen Horowitz. Other participants included Creator Ventures, SV Angel, Instagram co-founder Mike Krieger, Oculus co-founder Brendan Iribe, Deepmind and Inflection AI co-founder Mustafa Suleyman, and O’Reilly Media founder Tim O’Reilly.
A source familiar with the matter told TechCrunch that the funding valued ElevenLabs at $99 million post-money. CEO and co-founder Mati Staniszewski said the investment would support the company’s voice AI research hub and products for publishing, gaming, entertainment and conversational applications.
ElevenLabs was founded by Staniszewski, who previously worked at Palantir, and Piotr Dabkowski, an ex-Google employee and Staniszewski’s childhood friend. The founders were inspired by the dubbing of American movies they watched while growing up in Poland.
Projects targets long-form spoken content
The company’s platform turns text into speech using synthetic, cloned or newly created voices. Its language-agnostic models can also be fine-tuned by corporate customers to build proprietary speech models.
Alongside the funding announcement, ElevenLabs launched Projects, a workflow for creating and editing long-form audio. Users can create dialogue segments and audiobooks within the platform. Staniszewski described potential business uses such as multilingual audiobook production, character voices for video games, spoken versions of digital articles, access to online text for visually impaired people and AI radio.
ElevenLabs launched in beta in late January and gained attention for the quality and speed of its generated voices, as well as a generous free tier. The company said it had more than a million registered users across creative, entertainment and publishing fields, who had created ten years’ worth of audio content.
Detection tool addresses risks of voice cloning
The platform’s popularity also brought attention to misuse. The source reports that users on 4chan used ElevenLabs to create hateful messages imitating celebrities, including actor Emma Watson. The Verge’s James Vincent also demonstrated that voices could be cloned in seconds to produce audio containing threats and discriminatory language.
ElevenLabs said it would add safeguards, including restricting voice cloning to paid accounts, banning users who repeatedly break its terms of service, and offering an AI detection tool. The newly launched AI Speech Classifier is available as an API to “selected” partners. It is designed to identify whether an uploaded audio sample contains content generated by ElevenLabs.
The classifier addresses one part of the challenge: identifying synthetic audio. The source notes that a voluntary detection tool may not stop misuse, and raises a broader concern about the technology’s implications for voice actors. It reports that actors are increasingly being asked to give clients rights to use their voices for synthetic versions, potentially without extra compensation.
Expansion plans include dubbing
ElevenLabs has described work with Storytel, TheSoul Publishing and MNTN on audiobooks, video games and radio content. Storytel and TheSoul Publishing are strategic investors. The company also plans to extend its models to voice dubbing, aiming to carry emotion and intonation from one language into another while keeping the original speaker’s voice.
ElevenLabs said it was already testing AI dubbing at scale with industry partners. Its broader competition includes Amazon, Google and Microsoft, as well as Murf, Tavus, Resemble AI, Respeecher, PlayHT and Lovo. The company reported having $21 million in the bank, including $2 million from a pre-seed round in January.