Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders. Subscribe Now
The “click here to listen to this article” audio snippet that appears at the top of some web pages is extremely helpful for those with visual or reading comprehension problems — and, increasingly in today’s fast-paced world, those short on time.
Prolific voice AI startup ElevenLabs is looking to get in on this market with the launch this week of Audio Native. The embeddable audio player automatically narrates web page content via the company’s text-to-speech technology.
It is the latest in a continuous rollout of capabilities from ElevenLabs: Also this week, the two-year-old unicorn released ElevenLabs Reader, which can voice text from web pages and documents in 11 different voices. The company’s models speak 29 languages and can also dub full-length movies and turn prompts into song lyrics.
The new Audio Native is available at the “creator” tier for $11 a month, and the tool also features built-in metrics and a listener dashboard that allows users to track audience engagement.
On its X page (formerly Twitter), the company pointed to web pages that have utilized its capabilities, including its own blog, an intro to using AI for SEO from bensbites.com and a November 2023 New Yorker article, “Not all of America’s national-security threats are overseas.” ElevenLabs has also been used by The Atlantic and The New York Times.
“It’s customizable, easy to set up, and helps drive reader engagement while making your content accessible for readers (and listeners) around the world,” Sam Sklar of ElevenLabs wrote in a blog post.
Voicing websites with ‘brief snippets’ of HTML
Audio Native users can embed and automatically voice their site or embed audio from an existing project or via ElevenLabs’ API.
AI Scaling Hits Its Limits
Power caps, rising token costs, and inference delays are reshaping enterprise AI. Join our exclusive salon to discover how top teams are:
- Turning energy into a strategic advantage
- Architecting efficient inference for real throughput gains
- Unlocking competitive ROI with sustainable AI systems
Secure your spot to stay ahead: https://bit.ly/4mwGngO
To voice website content, they need to provide a “brief snippet” of HTML, according to ElevenLabs. They must add their website domain to the “allow” list, choose a voice (presumably from the company’s existing 11 personalities), customize their player’s background and text color and then copy and paste embedded code onto their website.
An optional pronunciation dictionary can specify the phrasing of words unique to a brand. The model will create voiceovers of all text content on a page by default, but this can be customized with CSS selectors.
The new capability currently supports React, Squarespace, WordPress, Ghost, Webflow and Framer.
Early users are calling the tool “sick” and “amazing,” and others are touting its ability to help improve accessibility.
Presumably based on its social posts, ElevenLabs intends to continue its capability rollout spree: In a thread announcing Audio Native on X, one user asked: “Give us RSS feeds? Then we can make a podcast out of our written content ?”
To this Luke Harries, head of growth at ElevenLabs replied: “Great idea, sharing with the team.”
ElevenLabs, which was founded in 2022 and claims a valuation of $1.1 billion, was founded by former Google engineer Piotr Dabkowski and former Palantir Technologies deployment strategist Mati Staniszewski. Its most recent funding round was $80 million in January.
The company is innovating in an increasingly competitive market alongside such players as Speechify, Deepgram, Voicemod, Murf, LiSTNR, and LOVO. But there is no doubt ample opportunity, as the global AI voice cloning market size is expected to reach $16.2 billion by 2032, representing a compound annual growth rate (CAGR) of nearly 28% from 2023.
Notably, ElevenLabs has teamed with HarperCollins Publishers to produce AI-generated audiobooks, and it has also launched a marketplace where users can sell their cloned voice for money. However, the company has also come under scrutiny, notably when it comes to its music generation capabilities — specifically, whether it has used copyrighted materials to train its models, an increasingly contentious topic of late.
