Posted June 20, 2023

Listen to Bryan Kim read this blog post, with voice generated by ElevenLabs.

The human voice is one of our most primal, personal and expressive forms of communication. From a newborn’s first babbles to Martin Luther King Jr.’s “I Have a Dream” speech, voice carries not only our ideas, but also the most profound emotions and connections.

To date though, it’s been a time intensive and costly challenge for developers and creatives to incorporate high-quality voice into their work. ElevenLabs is on a mission to change this, by giving every program a voice through its proprietary speech synthesis, voice design and cloning technology.

While text-to-speech (TTS) technology has been around for decades, it hasn’t been able to reach its full potential–most synthetic voices lack engaging intonations and enunciations, offering no sense of personality. And though premium voice services are available, lengthy production processes and high costs have kept the technology out of reach for most real-time and interactive use cases. That’s why podcasters still require expensive equipment and quiet rooms to record every sentence, and game designers can only afford voice actors for main characters, leaving nonplayer characters silent. 

Since ElevenLabs first launched its voice design and cloning product in January, it has substantially improved upon existing TTS models while making the technology significantly more accessible. It quickly became a go-to tool for creators and developers across media, gaming, and content creation industries. With a few clicks, their state-of-the-art foundation model is able to generate voices that sound incredibly human, with proper pause, intonation, and breathing rhythms. You can even create a clone of your own voice from a 30-second clip. Since launch, ElevenLabs has amassed over 1 million registered users who have generated over 10 years worth of audio content.

Already, ElevenLabs has been embraced by a broad, creative community. The technology has enabled indie authors to create audiobooks, assisted developers to voice secondary characters in video games, helped the visually impaired access online written content, and even led to the creation of totally novel experiences, including entire podcasts narrated with AI, like Seth Godin’s Akimbo, and fully generated mini films. Companies from publishing (like Storytel) to gaming (like Paradox Interactive) are already adopting ElevenLabs in their work. Once the creative floodgate is open, we’re left in awe of this technology. Imagine the possibilities unlocked when creatives can give any character any voice, and have that voice be indistinguishable from the original.

Listen to Jennifer Li read this blog post, with voice generated by ElevenLabs.

What’s more exciting is the fact that these voices are inherently multilingual. ElevenLabs’ model supports text-to-speech conversion across French, German, Hindi, Italian, Polish, Portuguese, and Spanish (with more to come). In a not too distant future, we’ll be able to listen to our favorite podcast or watch our favorite movie in our native language, while retaining the actor’s original voice.

This future is very personal to ElevenLabs’ founders, Mati Staniszewski and Piotr Dabkowski, who grew up in Poland and were often frustrated by the poor dubbing of American movies. Armed with deep domain knowledge from Google and Palantir building ML products and large scale systems, today the founders and their team are determined to break down linguistic barriers and bring the power of voice to any program or platform. 

At a16z, we are strong believers that generative AI tools will revolutionize the creative suite–unlocking more creativity by enabling professionals to produce more content with higher quality and empowering a wide swath of net new creators thanks to more accessible, intuitive tools. We are thrilled to be joining the ElevenLabs board and co-leading their Series A alongside Nat Friedman and Daniel Gross.

Listen to Justine Moore read this blog post, with voice generated by ElevenLabs.