Eleven v4: Voice cloning this fast should not sound this good

Eleven v4, the latest voice model from ElevenLabs, is available now in BeyondWords.

As someone who spends a lot of time working with AI voices, I can tell you it represents a striking leap in quality.

In my testing, v4 voices read out articles much like a professional narrator would. Prosody is natural with emphases placed to give the correct meaning to each sentence. Delivery is context aware, so the article content has real feeling, and longer pieces maintain their flow and character much more consistently from one section to the next.

This ElevenLabs model also supports IPA transcriptions for finer control over custom pronunciations, along with a wider range of languages and accents.

But instant voice cloning is where things really get interesting.

More faithful clones in just 10 seconds

I’ve found that Eleven v4 produces significantly more accurate instant voice clones than previous models. With as little as 10 seconds of recorded speech, it can capture far more of a speaker’s accent, delivery, and individual style. 

You can listen to my own v4 instant voice clone narrating this article.

I also asked my brother, Chris Piercy, to record a short audio sample for me. Here, you can compare his original recording with his v4 instant voice clone. 

If you ask me, they’re hard to tell apart.

For publishers, this level of fidelity changes what instant voice cloning can realistically be used for. It becomes much easier to create audio around the individual voices, accents, and styles that already exist within a newsroom.

Give every byline a voice

Instant voice cloning only requires a short, approved audio sample and is included for free in every BeyondWords subscription. Once generated, v4 clones can be applied to the articles you choose. 

So, every story can be automatically narrated by the person who wrote it, in a voice that sounds more realistic than ever.

For podcast hosts, broadcasters, and other journalists already known by voice, this adds instant familiarity and personality to audio content. Regional publishers can preserve local accents that make listening experiences feel much more authentic to the communities they serve.

Better voice cloning also gives publishers another way to build a distinctive audio identity and create richer experiences that feel recognizably their own.

Voice is increasingly becoming part of a publisher’s identity, not just a way of narrating an article. As audiences spend more time with audio, headphones, speakers, and voice interfaces, the voices they associate with a publisher will matter more.

Publishers like La Nación are already proving the concept with our highest-quality professional voice clones. With v4 in BeyondWords, bringing this kind of experience to more newsrooms becomes much easier.

Try v4 in two minutes

Already a BeyondWords publisher? You can create an instant voice clone and start testing it with v4 in just a couple of minutes. Go to Custom voices in your organization dashboard, click + Voice, and follow the steps. Once your clone is ready, select the v4 model and head to the Editor to start testing. 

You can also try v4 with premade and generated voices, or talk to our team about creating an even higher-quality v4 professional voice clone.

Not yet a BeyondWords publisher? Book a demo to see how you can keep up with ElevenLabs’ latest voice models through a platform built for publishers.