BeyondWords Blog

At BeyondWords, we care about audio quality. It's our north star - the single greatest driver of engagement through audio.

Today's audiences are discerning. A single mispronounced acronym, misread measurement, or garbled foreign word can break the spell and break trust.

That's why we're excited to introduce AI preprocessing - a small step for AI audio, a giant leap for listenability.

Read more on this, plus hear about some sleek new updates to the Editor, below.

Prepare for cleaner audio, at scale

Even the most realistic-sounding AI voices can be let down by mispronunciations.

When a sports score is read like a year or a date sounds like a price, it can pull listeners out of the story and remind them they're not hearing a human.

AI preprocessing fixes this automatically. It understands the context of your article and normalizes tricky text - like dates, sports scores, measurements, and abbreviations - into smooth, natural speech.

With built-in language detection, pronunciations of foreign words within articles can now also be automatically converted into their native pronunciations.

Switch on AI preprocessing today in your Projects preferences or directly in the Editor.

Make room for the new Editor

We've rebuilt the Editor from the ground up, making it faster, cleaner, and more powerful.

Clone a voice without leaving the Editor, assign it to your article, and adjust voice, summarization, and video settings on a per-article basis.

What's more, you can now import any article straight from a URL to speed up your workflow.

Visit our docs and guides for more info.

Fixes and improvements

We've made several behind-the-scenes upgrades to improve flexibility, compatibility, and security:

Video: Adjust how many caption lines appear per scene, control caption alignment, and define how article images display and playback in videos.

Magic Embed: Now works reliably with sites that render articles dynamically via JavaScript.

SAML: Added support for SAML Single Sign-On (SSO), making logins more secure and streamlined for enterprise customers.

If you have any questions about these updates, just reply. We're always happy to help.

That's all from us for now. If you're heading to INMA Media Innovation Week this September in Dublin, we'll be there! Come say hello!

And to keep up with the most recent updates, visit the BeyondWords changelog.

Not all text is designed for audio. Think sport scores, stock market updates, analysis of demographic data, etc.

For news publishers, even a single mispronounced name, clumsy acronym, or mistranslated currency can damage trust. But editing for audio can be tedious and impractical at scale.

That's why we've launched AI preprocessing - a new feature designed to make AI audio smarter, clearer, and more natural-sounding from the very first listen.

Have confidence in every article

Priming text for multimodal formats has always been a focus for BeyondWords.

While AI voices sound more natural than ever, they can still be tripped up by the structure of real-world editorial content.

BeyondWords offers publishers full control over audio and video with pronunciation rules - such as substitutions and IPA transcriptions - embedded directly into the dashboard.

With AI preprocessing, we take this a further step forward, removing the need for subediting and offering confidence in every article as you scale.

Reduce time spent on manual improvements

AI preprocessing automatically detects and cleans up non-standard text - such as dates, scores, and abbreviations - removing the need for manual rules.

Take this sentence:

Snap rose 24.3% YoY to 422M, contributing to 8.9% QoQ revenue growth in 3Q.

Before AI preprocessing:

After AI preprocessing:

In this example, numerous terms are preprocessed to sound natural in spoken English:

"YoY" becomes "year over year"

"QoQ" becomes "quarter over quarter"

"3Q" becomes "third quarter"

"422M" becomes "four hundred twenty-two million"

Or take this Spanish sentence:

El Gobierno anunció un subsidio de $150.000 para las familias afectadas por la sequía en la región de Valparaíso.

Before AI preprocessing:

After AI preprocessing:

Here, thanks to context-aware AI preprocessing, the "$" symbol is correctly interpreted as referring to Chilean pesos, not US dollars.

Perfect pronunciations of foreign names

Furthermore, AI preprocessing allows for language detection for improved pronunciations of foreign words.

Currently available in German, with more languages coming soon, this improvement ensures that named entities, quotes, and loanwords are spoken in the correct accent and intonation - no clunky mispronunciations or awkward phrasing.

For example, take this German sentence:

Im neuesten Update kündigte das Start-up ein Rebranding seiner App an, das laut CEO mehr User Engagement und eine bessere Conversion Rate bringen soll.

Before AI preprocessing:

After AI preprocessing:

In this sentence, "Start-up," "CEO," "User Engagement," and "Conversion Rate" are all detected as English terms and pronounced accordingly.

It's these little touches that make a big difference. And now, they happen automatically.

Why AI preprocessing matters

With AI preprocessing, your audio gets cleaner and more consistent, without any extra effort from your team.

Fewer manual rules

No more fiddling with exceptions like "12/03/24" or "1,500 CAD". AI preprocessing handles edge cases in context, so you can skip the tweaks and focus on what matters.

Professional grade output

A clumsy misread breaks immersion and trust. By getting pronunciation right the first time, your audio sounds smooth, confident, and credible.

Multilingual support at scale

Ideal for global newsrooms, foreign names and terms are detected and pronounced correctly.

Whether you're scaling audio across hundreds of articles or experimenting with article-to-video, AI preprocessing helps every article feel polished, at scale.

How to turn AI preprocessing on

Now available in public beta, follow these steps to turn on AI preprocessing:

1. Within your Project, head to "Content" then "Preferences"

2. From the top navigation, click the "Pronunciations" tab

3. Click the "AI preprocessing" button

4. Toggle the switch on, then click "Done"

To enable AI processing for a specific article, open the article editor, go to the "Pronunciations" tab in the sidebar, toggle on "AI preprocessing", then gene...

In a world flooded with news, how do you make sure your stories stand out and get heard?

At BeyondWords, we've got a few audio tricks up our sleeves to help your content rise above the noise.

Introducing Summaries, Playlists, and Podcast Feeds:

AI summaries

Our AI-powered summaries automatically transform your articles into concise, digestible audio snippets. Perfect for listeners on the go who want the headlines and highlights… right when they need them.

Custom playlists

Create smart or standard playlists that organize your audio content around themes, trending topics, or audience interests. Whether your listeners prefer deep dives or quick catch-ups, they'll find exactly what they're looking for.

Podcast feeds

Turn your audio into ready-to-go podcast feeds that work with all the big platforms, including Spotify, Apple Podcasts, and YouTube. Expand your reach and let listeners subscribe, download, and enjoy your stories wherever and whenever they want.

Together, these tools don't just amplify your voice: they make sure it's heard loud, clear, and exactly where your audience is listening.

Want to know more? Feel free to reach out to one of our team at [email protected].

Don't you love it when you time your commute just right, when you're first in line at the coffee shop, when you meet your deadlines…when things just work out?

Well, we've been trying to package up that feeling at BeyondWords with the new Magic Embed.

Forget tedious manual processes, complicated coding, or juggling multiple tools. Easily embed BeyondWords and make any article available in audio.

Here's how it works:

1. Paste and play: Configure your settings, paste the embed code into your page, and voilà! Magic Embed does the heavy lifting - automatically generating, embedding, and updating audio.

2. Stay synced: Edits happen. Stories evolve. Magic Embed keeps your audio seamlessly in sync, updating instantly whenever your text changes. No lag, no fuss, just real-time magic.

3. Engage effortlessly: Our player is sleek, accessible, and responsive, effortlessly fitting your brand's look and feel. Readers can listen anywhere, on any device.

Instead of wrestling with complex integrations, you now have more time to focus on what really matters: producing amazing stories that resonate with your audience.

More importantly, you've given your readers the freedom to explore what matters most to them, right through their headphones.

Want to know more? Feel free to reach out to one of our team at [email protected].

We're thrilled to announce our partnership with Stavanger Aftenblad, one of Norway's oldest and most well respected regional newspapers.

The new partnership marks an extension to our ongoing commitment to the Schibsted Group, as we work together to bring state-of-the-art AI audio to their esteemed titles.

For Stavanger Aftenblad, preserving its strong regional voice was essential. They chose the voice of journalist Ola Myrsef as the voice of their articles.

Elin Stueland, Head of AI, says: "Audio articles have been an exciting addition for Stavanger Aftenblad, allowing us to make our journalism even more accessible and engaging for our readers. The feedback has been very positive, with our audience appreciating the flexibility to consume content on the go.

"We're also thrilled that the technology is able to pronounce local names correctly, which is particularly important to us as a regional newspaper that is deeply connected to our readers and the local community."

At BeyondWords, we're proud to help Stavanger Aftenblad connect their readers with the news that matters most to them, allowing them to reach new heights of their storytelling.

In 2024, BeyondWords helped leading Argentinian news publisher La Nación to clone the voices of 13 of their most respected journalists.

It was for a new feature designed to keep subscribers hooked. A chance for readers to hear news read by their favorite journalists. A way to bring the spirit of La Nación's newsroom to people's ears.

A great time to innovate

We had worked with La Nación for some time already, using one of our premade Argentinian voices to turn their articles into captivating audio. In June 2024, their product team chose to crank those articles up to eleven.

They were looking to capture the imaginations of their readers and their newsroom. And they were determined to be ambitious-by embracing every possibility of AI audio.

"This wasn't [just] about engagement. This was about innovation," Fernando Silva, Lead Product Owner at La Nación, told us.

We know that listening to familiar voices improves how people connect with and understand what's being said. In collaboration, we decided to test that insight.

The team planned to replace their existing premade voice with a number of voices based on their journalists, using our Professional voice cloning service. They chose the journalists who would read a script we'd created from their articles. We needed to get them excited.

Convincing the newsroom

It was crucial that La Nación's journalists saw the possibilities of this project. Audio articles infused with their own personalities.

For many though, it felt like a giant leap into the unknown. How would the voices sound? Would time in the recording studio be well spent? How would this "experiment" work in practice?

They needn't have worried. The voices were met with a chorus of positive feedback from the newsroom, and colleagues were soon clamoring to have their own voices cloned.

"Once we heard the final voice, it was easy to convince the rest of the journalists," added Fernando.

Listen to Carlos Pagni, renowned Argentinian journalist, below:

In the end, we trained and launched 13 voices with La Nación-each alive with individual character and resonance.

The moment of truth

The new voices were launched to La Nación's subscribers in early December 2024 to a unanimous thumbs up.

"Not only did we receive a positive reaction from the journalist community, our readers have told us it was an excellent initiative," Fernando shared with us.

Cloning the voices of journalists has both improved how articles speak to subscribers, and given different audiences a more diverse listening experience, boosting engagement with the brand.

And La Nación aren't alone. More publishers are integrating audio articles into their subscription products and paywalls. In an age of information overload, it's a proven way to capture attention and keep subscribers happy.

Explore with us

As voice cloning becomes more plug-and-play, we're excited about the possibilities of connecting publishers with their readers through the personality and character of their own newsrooms.

Keen to learn more? We're always happy to chat. Email us at [email protected]. Or reach out and book a demo.

Have you ever cringed at the mispronunciation of a key name or location in an article? Whether narrated by a human or an AI voice, getting uncommon words right can be surprisingly tricky.

From Scottish towns like Hawick (pronounced "Hoik") to Iceland's infamous Eyjafjallajökull volcano, the stakes are high when missteps could alienate your audience - or worse, prompt an avalanche of corrections.

For news publishers relying on AI-generated audio, this challenge becomes even more daunting. Correct pronunciation is essential to maintaining credibility and trust, and AI voices depend on accurate phonetic guidance to get it right.

Enter Magic IPA

At BeyondWords, we're obsessed with quality. For us, that means your audience is talking about your content - not distracted by mispronunciations. Since our inception, we've allowed customers to adjust pronunciations using the Rules feature in the BeyondWords dashboard. But we knew we could make it even easier.

This week, we introduced Magic IPA, a new feature designed to simplify and accelerate how you correct mispronunciations.

Here's how it works:

1. Open the Editor for your article

2. Find the problematic word and highlight it

3. Click "Add pronunciation rule"

4. Select "IPA transcription" in the "Type" dropdown

5. Click the magic wand icon next to the "Value" textbox

6. Suspenseful pause for a sprinkle of LLM magic…

7. "Preview" the new pronunciation

8. And finally, save your changes and regenerate the audio article

And voilà! Instant phonetic transcriptions for new or out-of-vocabulary words thanks to the power of LLMs.

Harnessing an LLM with some clever linguistic prompting, Magic IPA offers an easy, efficient way to refine your AI audio. While it's not always perfect (hence the name), it provides a reliable starting point, saving you and your team time and effort.

Side note, we'd love to hear your thoughts about this new feature. If you encounter any issues or have feedback, please reach out to us at [email protected].

We're thrilled to announce our latest partnership with Kleine Zeitung, one of Austria's most influential and widely read news titles. The collaboration introduces audio articles as a subscription-only feature, read in a voice familiar to their readers.

Listen to short demo below:

After developing a custom script from their articles, our team cloned the voice of Barbara Haas, the newspaper's Head of Podcast and Video. The result is a voice that maintains the journalistic integrity and personality readers associate with the Kleine Zeitung brand.

By offering an option to listen, Kleine Zeitung creates opportunities for their subscribers to connect - whilst commuting, exercising, or whenever away from their screens. Audio not only enhances accessibility but promises deeper engagement with subscribers, improving retention through a richer and more personalized experience.

Sebastian Krause, Head of Product at Kleine Zeitung, shared his enthusiasm for the partnership:

"This collaboration with BeyondWords is a natural step forward for Kleine Zeitung. By integrating AI technology and Barbara Haas' trusted voice into our offerings, we're providing our subscribers with a novel way to engage with our content. It's about blending innovation with tradition to keep our readers at the heart of everything we do."

At BeyondWords, we connect publishers and their audiences through the possibilities of AI voice. We help publishers cut through the noise and give people freedom to explore what matters most to them.

About Kleine Zeitung

Founded in 1904, Kleine Zeitung is Austria's second-largest regional newspaper, with roots in Styria and Carinthia. Renowned for its commitment to quality journalism, the newspaper offers comprehensive coverage across politics, culture, business, and local affairs.

With a combined print and digital readership that extends well beyond Austria's borders, Kleine Zeitung is a trusted source of information for hundreds of thousands of loyal readers. In recent years, it has embraced digital innovation to complement its traditional print offerings, solidifying its position as a forward-thinking leader in the media landscape.

We're thrilled to have partnered with the fantastic teams at Ringier Medien Schweiz, Die Presse, and La Nación to bring audio to their articles.

These collaborations bring audio to millions of readers and subscribers globally, giving the flexibility to listen whenever they don't have time to read, all in the authentic voices of their talented journalists and editors.

Professional voice cloning

We've made significant improvements to our voice models in recent months, improving naturalness, whilst significantly reducing data requirements and training times.

We're also excited to announce professional voice cloning support for 84 languages and accents, including:

Afrikaans, Arabic, Bulgarian, Catalan, Chinese (Cantonese), Chinese (Mandarin), Czech, Danish, German, Greek, English, Spanish, Finnish, French, Hebrew, Hindi, Croatian, Hungarian, Indonesian, Italian, Japanese, Korean, Malay, Norwegian, Dutch, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Swedish, Tamil, Telugu, Thai, Turkish, and Vietnamese.

Interested in cloning your own voices? Speak to us.

Fixes and improvements

Resolved a bug that caused duplicate audio segments when regenerating an audio article with a new voice.

Fixed an issue that prevented some users from uploading audio.

Fixed a bug where unpublished audio was mistakenly published on WordPress despite automatic publishing being disabled.

Deprecated X-Sampa support in favor of IPA for phonetic spellings.

Deployed a fix to prevent duplicate subscriptions.

If you have any questions about these updates, please don't hesitate to reach out!

For the most recent updates, visit our dynamic changelog.

Thomson Reuters is a global leader in providing services and information to industries that require accuracy, speed, and reliability.

In early 2024, they partnered with BeyondWords to transform how their audiences engage with their legal directories, FindLaw.com and Abogado.com.

In this post, learn how Thomson Reuters leveraged AI voices to enrich interactions with their customers, drive engagement, search engine performance and accessibility.

Improving reader engagement with new formats

In recent years, as attention has moved away from primarily text-based articles to content that incorporates audio and video, publishers globally have had to adapt.

The ability to listen has become particularly appealing as we spend more time multitasking and wearing headphones. In fact, between 2018 to 2023, daily listeners of spoken audio increased by a huge 17% in the US.

The team at Thomson Reuters developed a strategy in response to these changing habits.

"Who wants to read?!" said Amelie Collins, Product Innovation Manager at Thomson Reuters. "Audio makes our articles more appealing, especially for busy professionals who might not have time to sit and read. It's a practical way to meet our audience where they are-whether they're commuting, exercising, or multitasking."

Since launching audio articles on Abogado.com and FindLaw.com, the results have been overwhelmingly positive.

On average, listeners on Abogado.com engage for 3 minutes and 6 seconds-approximately 4x the average time spent with the average news article.

"More than 80% of our Abogado.com visitors are using mobile devices. The mobile friendly BeyondWords audio player allows our users to engage more with our site while they are on the go." Collins said.

Audio for SEO

To drive traffic to their legal directories, Thomson Reuters relies on search engine optimization (SEO).

They recognized that offering an alternative way for their readers to engage with articles would boost overall dwell time-a factor closely tied to better search engine performance.

In a recent study, we found that listeners are 32% more likely to engage in multiple sessions than non-listeners. Furthermore, listeners visit 19% more pages than non-listeners.

Improved average time-spent and lower bounce rates send positive signals to search engines helping to improve page rankings.

"We've seen double digit percentage increases in clicks and visibility from search engines and return visits to content where we have added the BeyondWords player. More surprising, we've seen over a 200% increase in organic search engine traffic and visibility since we have added the audio player to all of our blog posts."

Audio can also help reach a wider audience of those who prefer to listen on the go, further increasing site traffic.

Essential accessibility standards

Lastly, for Thomson Reuters, ensuring the highest level of access to legal insights is a top priority. Access to accurate legal information is essential for professionals to provide sound advice and empowers consumers to make informed decisions about their rights and responsibilities.

To ensure that their insights are accessible to all, Thomson Reuters adheres to the Web Content Accessibility Guidelines (WCAG) 2.1 standards.

Adding audio was essential to maintaining these standards and confirming the BeyondWords Player met the specific functionality was a key acceptance criterion.

To achieve this, our teams collaborated closely to enhance the player's functionality, upgrading keyboard controls to ensure full WCAG 2.1 compliance.

Collins added: "At BeyondWords questions and problems are not only answered and solved, they are welcomed in the true spirit of improving the product."

We're always eager for the opportunity to collaborate more closely with our customers-not only to solve issues but to refine and enhance our product continually to better meet their evolving needs.

Want to know what we could do for you? Book a meeting with our team.