SpeechKit Audio Articles

The speech services market is young but already being disrupted. How do you compare company X or company Y in such a fast moving environment?

In this post, we lay out how we see this rapidly evolving market and what differentiates one company from another. First, some background...

Text-to-speech has often been overlooked as a revolutionary new media format. Over recent years, as with much nascent AI, text-to-speech has been scrutinised over error rates, poor voice quality, and its ‘robotic-ness’. After 4-years of working to improve this technology, listening through hours upon hours of audio data, our team is well versed in these imperfections.

However, great improvements have been made - by our team and by our peers. Last year, Nic Newman at the Reuters Institute predicted audio articles would become ‘standard’. New neural voices from SpeechKit, Amazon, and Google, among others, are coming closer to human speech than could have ever been imagined. Furthermore, text-to-speech is starting to show very real results. The average engagement time of a listener is more than four times that of a non-listener .

Text-to-speech is being adopted to improve customer value and fast-track digital transformation. Below we’ve segmented the market into three categories to make for easier comparison.

Cloud Service APIs

Services, such as Amazon Polly , Google WaveNet and IBM Watson , provide text-to-speech APIs as part of their cloud service platforms. These products have been developed over the last 10-15 years, producing some of the most robust voices on the market. Early adopters, such as Bloomberg and The Globe and Mail , adopted Amazon Polly across their publishing to give their readers the option to listen. Whilst high-quality, these services lack utility and customization; they’ve produced very powerful APIs, yet haven’t created any tools (hosting, CMS integrations, analytics, monetization) to help publishers get the most from text-to-speech.

Strengths: reliability, voice-quality, cost.

Weaknesses: ease-of-integration, customization, publishing tools.

Custom Voice Startups

Speech startups, such as Resemble and Sonantic , aim to out-compete those cloud services mentioned above, by creating custom synthetic voices. Similarly, these companies have deep expertise in Machine Learning, developing high-quality voices for sectors such as gaming and customer service. Similarly to cloud services, whilst these companies have developed advanced APIs, they’ve neglected publishing tools. Neither of these examples provides audio distribution tools, analytics or monetization options for publishing at scale.

Strengths: voice-quality, customization, ease-of-integration.

Weaknesses: publishing tools, cost, scalability.

Text-To-Speech Startups

Startups, such as Trinity Audio and Play.ht , are our most direct competitors. These, like ourselves, are developing tools around text-to-speech, creating value for publishers and bloggers. Trinity Audio have focused on monetizing APIs created by those cloud services mentioned above. They’ve built ad insertion technology and partnered with multiple audio DSPs to improve fill-rates and potential CPMs. Their business model aims to provide revenue back to publishers who adopt their tech, while they take a cut. Whilst this is sound in principle, we believe a focus on voice quality, infrastructure and customer service most benefit our customer’s long term.

Strengths: monetization, cost, scalability.

Weaknesses: ease-of-integration, voice-quality, customization.

SpeechKit

In comparison, SpeechKit has focused on becoming the ‘full-stack’ service for automated audio publishing. We help news publishers, institutions, corporate businesses and bloggers publish their written content in audio, automatically and at scale. We believe that improving access to quality content, through seamless audio integrations, is the real power of text-to-speech. Our strengths lie in building scalable products, using domain-specific voices , tha...

Text-to-speech audio articles are growing in popularity worldwide. How will new localized accents and dialects improve engagement?

As reported by the 2020 Spoken Audio Report , produced by NPR and Edison Research earlier this month, we are listening to more spoken audio than ever before. Spoken audio's share of listening has grown 30% over the past 6 years, 8% this last year.

Text-to-speech technology has come a long way in the last 6 years. Gone are those days when interacting with automated voices was confined to navigation, service centres and buggy early AI-assistants. Text-to-speech has proliferated into 50+ languages, and many more different voices, tones and styles .

SpeechKit works primarily with news publishers who, until now, have been limited to adopting off-the-shelf voices, provided by Amazon Polly and others. These voices have worked great to increase engagement with news articles and provided readers with an option to listen to audio articles.

However, there’s room for improvement.

Cape Town, home of News24's local voice. Image credit Zoë Reeve .

Resonance keeps us engaged

All brands, whether that be in news media, consumer or any other sector, want to talk to their audience in a voice that resonates. This is the sweet spot of accent, intonation and style which satisfies listeners. Its why great voice artists can charge what they do and (what we’re finding at SpeechKit) it’s what keeps listeners engaged with audio articles.

Studies show that listening comprehension improves when a sentence is spoken in someone’s native accent. We’re more receptive to voices with meaning to us. Having lived much of my childhood in Australia, I live in the UK full-time now, yet my Siri is programmed to speak to me in an Australian accent. For one reason or another, it makes that experience personal – it gives it context and meaning. I find myself enjoying listening to the responses to my (less than meaningful) requests.

Custom localized synthetic voices

We’ve been helping publishers adopt audio articles for the past 3 years and we’re making good progress. Most recently we’ve started to develop custom synthetic voices for our customers who hope to create a voice that better resonates with their audience. Our first publisher, News24 , South Africa’s largest news platform needed a South African-accented voice that could pronounce local names in Zulu, Xhosa and Afrikaans.

Quite like this new News24 feature where you can listen to articles. For all sorts of reasons, it's a mighty step forward. — Bhavna Maharaj (@BhavMaharaj) August 13, 2020

We developed the voice using a new machine learning technique to model realistic voices, requiring less training data than conventional methods. The voice was launched as a premium feature on a new digital subscription product at the beginning of August. During that month we observed an 407% increase in audio engagement on News24.com when compared to publishers using conventional text-to-speech voices. Furthermore, over the past two months average listen length per article has increased to 2 minutes and 12 seconds.

We’re confident that custom voices, tuned to the ear of the audience, are going to introduce new growth to text-to-speech audio articles. Advances in voice training are providing the efficiencies needed to allow for specialization and providing publishers the quality they need. Technology is allowing us to address the subtleties of voice, encouraging the further adoption of audio articles.

More about SpeechKit

At SpeechKit we’re helping 100’s of news publishers to automate audio versions of their news articles. Sign up for a free trial and instantly start engaging the audio generation.

Our guiding principle has always been to improve text-to-speech integrations and to create audio for every story.

For the past 5 years, SpeechKit has been developing text-to-speech tools for digital publishers. Our guiding principle has always been to improve text-to-speech integrations and to create audio for every story. We want more than anything to create seamless audio experiences — that make listening to the content we crave effortless.

This week, we release a brand new minimal audio article player.

Designed to fit inline with existing publishing and branding, the sleek new audio player sits atop each article. We’ve designed it with customizable features for enterprise use — such as play and pause buttons, progress bar and new background colors. We’ve also added multi-speed listening to allow for faster or slower listening speeds.

Minimal design text-to-speech.

The new audio article player comes as default on new projects and we’ll be transferring existing customers over the coming weeks. We’ve outlined a number of updated features below. For more information on SpeechKits integrations please get into touch with someone on our team .

Player features

We designed the new player taking feedback from customers, minimising integration and design hassles. Once integrated using one of the SpeechKit integrations tools — WordPress , Ghost , API and RSS — the player instantly serves audio alongside your content.

CMS Integrations — The SpeechKit player integrates into your existing digital publishing tools for maximum efficiency. Once integrated, audio is published simultaneously alongside each piece of content. We’ve built integrations that fit the spectrum of digital publishing, including plugins for WordPress and Ghost , a SpeechKit API and RSS feed ingestion.

Audio Analytics Tracking — Impressions, listens and duration are tracked on every audio article. Customers of SpeechKit benefit from real-time tracking available through their project’s dashboard. Track trends and performance of audio articles as your readers turn to listening. Alternatively, integrate with Google Analytics to track audio within your existing GA dashboard.

Multi Playback Speed — We updated the player to include multi-speed playback. Speeds now available include x0.5, x1.0, x1.25, x1.5 and x2! Everyone’s got their style when it comes to listening. I’m a fan of the gentle and melodic x1.25!

HLS Adaptive Streaming — A priority for us at Speechkit has always been to optimize load speed and minimize the latency of audio article players. The new player includes HLS Adaptive Streaming, meaning audio quality adapts depending on the listener's available network data. This feature addresses on-the-go users moving in-and-out of good network as they go about their days.

JavaScript iFrame — Another speed-related design principal; reduce page loading speed. We never want our player to add to page loading, so we built the player in an iFrame which waits for the other elements on the page to load. Once it’s the turn of the iFrame the player loads in 110 milliseconds (as of last speed test), a 164% improvement on the existing player.

SpeechKit offers a range of audio content management tools alongside cutting edge text-to-speech technology. Digital content publishers use SpeechKit to distribute podcasts and other pre-recorded content, taking advantage of hosting and distribution options.

By harnessing the full capabilities of the SpeechKit platform, newsrooms, bloggers and agencies can manage all of their audio needs under one roof allowing for an easier and faster audio editorial process. This eliminates the need for multiple tools, platforms and dashboards.

In this post we'll outline some key features.

Players

SpeechKit audio players are designed to fit inline with web and mobile content. Our players work for both automated publishing (through WordPress, RSS or API) and/or with Mp3s uploaded directly into the dashboard. To get started set up a new project in the SpeechKit dashboard.

To set up a non-automated project - for podcasts or other content - create a new ‘plain-text’ project. Select a voice in case you choose to add text-to-speech articles too.

More info on setting up new projects in the docs.

Uploading Audio

Once a project is set up audio can be added using text-to-speech or by uploading pre-generated audio, podcasts or Mp3s. Hosting audio content on SpeechKit lets you distribute it through audio players and through other distribution channels - such as Spotify, Apple Podcasts, and Google.

In your new project, to upload a piece of audio content select ‘Add Article’, select ‘Upload Audio’, give your audio a name and hit ‘Upload Mp3 File’.

Distribute

From SpeechKit you can distribute audio to players embedded on your websites and apps, to Podcast Platforms and to Smart Speakers. Once you have audio you want to distribute head to the ‘Distribution’ tab in the dashboard.

If you want to embed a single piece of audio or Mp3 you can do this from the content tab. Click on the ‘< >’ icon next to the article name to get an embed code for that piece of audio content, or use the URL to embed into Medium and other sites.

From the distribution tab you can set up playlist players, feeds to Spotify and Apple Podcasts, as well as to Alexa and Google Home. Select a distribution option and follow the prompts to get started.

Check out this blog post for info on setting up feeds to Podcast Platforms.

Analytics

Lastly, using SpeechKit to publish audio will provide you access to data on how audio is performing. The analytics tab in the dashboard tracks listens, audio engagement, device engagement and more.

If you are a customer of Google Analytics you might also want integrate SpeechKit reporting with your GA dashboard. To integrate audio into your Google Analytics access the docs here.

The features of the SpeechKit’s audio content management system are designed to fit the needs of our publishing customers. If you have any question or feedback please contact [email protected].

Over the last few years, news of streaming wars between tech giants have distracted interest away from an interesting trend; in years since 2014, music streaming has decreased 5%, whilst spoken-word listening has increased by 20%.

As a population, we’re turning off the tunes in favour of information and stories. Audio is being utilised by a hungry, connected and agile modern media consumer. Furthermore, new audio formats that fit modern life are meeting this insatiable desire for information.

In the news industry, newsrooms around the world are ‘pivoting-to-audio’ both to engage new demographics and to improve engagement with existing ones. The recent high profile acquisitions of Audm by the New York Times, Gimlet and Anchor by Spotify, combined with the recent release of Audio articles to Apple News+ are clear markers of growing interest/momentum in spoken news audio.

In this post, we want to outline the drivers of change, the valuable benefits to newsrooms and what’s next to come.

Complete News Packages

Over the last decade, we’ve observed a trend away news snacking, clickbait and tweet-sized news. As a recent Twipe Mobile study reveals, even in those younger demographics most attuned to snacking, an equal portion of the audience prefer to consume their news in complete packages.

News readers who lend themselves to this behaviour tend to be lucrative sources of reader revenue. As a result, newsrooms and startups, such as Tortoise, are developing news products that reflect this trend back towards ‘slow news’.

Spoken word lends itself to this not-so-new behaviour. Now that sitting at the breakfast table with a print edition no longer fits modern news habits, audio is filling the gaps. In routines interrupted by subway doors, slow pedestrians and intermittent instant messaging, spoken audio offers a refuge.

As more text-based news is brought online (to audio), more personalisation and richer listening experiences are introduced, expect listening to take an even larger role. Perhaps that traditionally offered by radio.

We first heard evidence of audio’s subscriber benefits whilst listening to The Economist’s then Head of Strategic Product Development, Denise Law, describes the ‘unread guilt factor’. The Economist had recognised as early as 2007 that a primary driver of subscriber churn was the feeling of guilt brought on by the sight of a stack of unread editions.

For their solution they turned to audio articles, narrated by paid voice actors. This has been an immensely successful strategy for their publication along with others, such as Zetland, where the audio edition is more popular than the written one.

A strong ability to engage and retain newsreaders makes audio a valuable tool. From our own data, we find that audio articles are incredibly popular with return users — 48% of those who listen are returning.

Another recent Twipe Mobile study found that format variety is the main factor leading to a decision to subscribe, even when compared to editorial coverage. As the market for reader revenue becomes more competitive, listening will become ubiquitous with paid news subscriptions.

Valuable Demographics

Digital audio presents a valuable opportunity to engage a younger audience. In the UK, younger age groups are four times more likely to listen to podcasts than over 55s, as reported by Reuters. Newsrooms are turning to audio to reach the next generation of paying newsreaders.

Operations like the NYT’s ‘The Daily’ and the FT’s ‘News Briefing’ are being used to draw in younger news consumers on free platforms — building brand loyalty and securing future revenue.

We observe a similar trend in audio articles. From our own data, 65% of audio article listeners are 44 and under (compared to 49% of newsreaders). Furthermore, those in the 18–24 bracket are 51% more likely to engage with audio, when compared to the average reader.

Audio is a powerful tool when it comes to engaging younger demographic...

A new term has appeared in digital publishing over the past 12-months.

Audio Articles are just what their name describes; news articles in audio.

Over the past few years, demand for news content delivered on-the-go has exploded.

The number of people in the US listening to spoken word audio increased 20% in the last 5 years (during a time when music listening shrunk 5%).

Up until recently, podcasts seemed to be the only valid answer to rising consumer demand for audio content.

Yet podcasts deliver a set of unique challenges to newsrooms.

How do you effectively replicate quality news coverage in podcast format?

You may need to hire new talent to host a daily show, build a production studio, negotiate new distribution deals with the major podcasting platforms, find a suitable audio hosting provider and finally build an audience.

All this for an audio format that proves financially lucrative for just a select few, due to the high level of competition in the Podcast market and platform oversaturation in the national and world news podcast segment.

Audio News Articles provide a simple and effective alternative to conventional podcasts

This format, proving to be popular with many publishers, takes existing articles and converts these into playable audio files that are served alongside their written stories.

No need for new strategies, setup costs, or marketing.

The average completion rate for an audio story is 90%. Hakon Mosbech - Zetland (Nieman)

This new format provides readers with the option to listen to any article.

Competition for time has never been higher – and tools to reach news on-the-go are increasingly valuable. Furthermore, readers are finding this useful.

At SpeechKit, we’re seeing increases as high as 1200% in time spent on page when Audio Articles are implemented on news websites. Learn more here.

Audio Articles are proving a useful tool for publishers searching for increased engagement with their stories. With SpeechKit, publishers of any size can implement natural sounding audio within minutes.

To find out more, signup for a free 7-day trial, or get in touch with our team using the form below.

Podcasts are format à la mode in 2019. Young people are being drawn away from traditional linear programming towards on-the-go digital audio. Music streaming services have laid the technical groundwork for a new storytelling format to dominate in the coming years. Podcasts have gone from quirky garage-band style productions to million-dollar productions from brands such as the New York Times.

However, launching a new podcast can take some work. First, you should decide what theme you are going to focus on. How is your podcast going to set you apart from the mass of competing shows on the Podcast Store? Then you have to start thinking about production. How much time can you afford to invest in your podcast? And is this going to take away from precious time spent creating other content? Will it be a worthwhile channel for your audience and for your bottom line?

Text-to-speech changes all of this.

Using the latest text-to-speech, creators may take existing text content and convert this into seamless human-like audio. No need for studio time, podcast hosting, scripts and guests, instantly convert text into audio using the suit of audio automation tools. In fact, there are a few of tts benefits;

Fractional costs vs podcast production.

Instant creation from existing content.

Publish in multiple languages, within minutes.

Distribute content on websites, podcast platforms and smart speakers.

Track this audio content across multiple devices.

Optimise websites for audio search engine optimisation.

Stay ahead of the game and your audience better access to content.

We’re listening to more and more digital audio every year according to The Reuters Institute of News, media consumers are reaching out, more and more, to those sites that offer an on-the-go option to engage. And this is just the beginning. As text-to-speech becomes seamlessly integrated into the world around us and news audiences convert from reading to listening — consuming news read out loud will become instinctive.

Speech Kit provides creators with lightening fast text-to-speech automation tools. Implement on WordPress or set up a manual project and turn your favourite articles into podcasts using the URL text converting tool. Get your content ready with SpeechKit with these simple steps;

1. Head over to SpeechKit and start a free account. Free accounts have access to 5 articles a month. Buy a $10 package to take advantage of podcast distribution and analytics.

2. Setup a manual or WordPress project — if you’re using WordPress download the SpeechKit plugin and enter the Project ID and API credentials.

3. Start creating audio — either publisher as you would on WordPress and see audio appear atop every article, or insert text or article URLs into your manual project and convert text into audio in the SpeechKit dashboard.

4. Once you have had your project set up and are ready to distribute your audio as a podcast head to the distribution tab in the dashboard (this is a premium feature), click ‘add distribution’ and add a podcast feed.

5. Once you have this setup you can connect SpeechKit to Apple Podcast Connect, Spotify, or wherever else you listen to your podcasts. Every time you convert a new article on SpeechKit a podcast is pushed to your feed!

To get started with audio versions, read aloud, head to SpeechKit and start an account. We work with publishers large and small and have built an audio toolkit that works with modern publishing technology, which includes a WordPress plugin.

For any questions on this or anything else, reach out via [email protected] or get us on Twitter @SpeechKit_io.

Our friends at Amazon have released a batch of new features.

SpeechKit utilises the latest text-to-speech technology to power NewsNet, our processing engine that transforms news content to produce the best audio available to publishers. Last week, Amazon released two new features to Polly, the text-to-speech division of the company. These voices are live on the SpeechKit dashboard.

Neural Text To Speech

The first of the two features, Neutral-Text-To-Speech (NTTS), has been applied to all 3 UK and 8 US voices on the platform and is described as improving the “naturalness and expressiveness” of these voices. Check out some examples below.

The second of the two features is an option only available through the US English voices, Joanna and Matthew. Newsreader style is applied to these voices to give them an edge and allows the listener to know when their listening into a piece of news by mimicking a traditional news anchor’s pitch and intonation. Check out the examples below.

Get Set Up

To enable these features in your SpeechKit project head to the project settings by clicking on the 3 vertical dots next to the project name (when hovered over). From the settings page, select one of the NTTS new voices; Amy, Emma and Brian (UK), and Ivy, Joanna, Kendra, Kimberly, Salli, Joey, Justin and Matthew (US). To enable one of the new Newsreader Style voices just select

More about SpeechKit

At SpeechKit we’re helping 100s of news publishers to automate audio versions of their content efforts. Sign up for a free trial and instantly start creating audio for the next generation of news consumer.

Improving text-to-speech with automatic SSML tagging

In 2018, 24.8 million people in the UK (45.5%) listened to online audio. Our belief is that this number will continue to grow and with it the percentage of newsreaders who will listen to audio narratives, such as news articles.

SpeechKit was designed to help news publishers pivot-to-audio, via audio versions of their news stories, without the time and cost required to narrate them — providing newsreaders with the choice of listening to news articles when reading is undesirable, boosting engagement with audio-streaming native demographics.

To keep costs low and the audio scalable we’re using the newly released, neural voices, available through the Amazon text-to-speech (TTS) service (Amazon Polly), to generate lifelike audio.

However for news brands seeking to use TTS, like Amazon Polly, to deliver the best audio experience they can, at scale, they’ll need to use SSML (speech-synthesis-markup) tags. Not using SSML will, without a doubt, lead to a sub-par audio experience and dissatisfied listeners.

SSML gives you additional control over how Amazon generates speech from text. Enhancing audio with SSML involves inserting specific tags into the text. Doing this manually for a single news article can take time, doing so for all published articles is almost impossible.

Amazon Polly supports SSML tags (see table below), but the service does not insert them for you. This usually requires context and Amazon did not develop Amazon Polly just for the news industry.

We’ve developed a middle-layer, called NewsNet, that, amongst other things, automates the SSML tagging process for news articles using a combination of rule-based and neural-network-based techniques.

This post will demonstrate the importance of using SSML when it comes to converting news articles into audio, and highlight the benefit to publishers of using SpeechKit to automate this process.

Amazon Polly accepts inputs as either plain text or SSML. For publishers using SpeechKit, NewsNet automatically cleans and converts all plain text from news articles into SSML and encloses SSML tags around paragraphs, sentences, specific words and phrases, amongst a few other things we’ll discuss in another post.

,

The first step is to wrap the text into a tag. This tells Amazon Polly to process the input as SSML. The second step is to indicate to Amazon Polly that the text should be read as a news item using the tag. The third step, and this is where NewsNet starts to shine, is to tokenise all words, sentences and phrases in the text and apply specific SSML tags to them using either our hardcoded rules or neural nets.

The first of these is the

and tags that indicate whether a string of text is a sentence or paragraph to ensure that appropriate pauses are inserted into the speech — periods are not always reliable segmentation points in news stories.

Other SSML tags inserted using NewsNet include, but are not limited to, , , , and tags.

In some cases, Amazon Polly struggles to pronounce specific words. This is quite common with brands, or in the case below with the president of South Africa. Over time we’ve added hundreds of words, common in the news, and their corrected pronunciation, to NewsNet so that they are detected, tagged, with the phoneme tag, and pronounced correctly appropriately.

,

Different news domains use different acronyms and abbreviations, that when spoken might sound unusual. NewsNet detects them, and using SSML, instructs Amazon Polly to expand them into their full spoken form using the tag or the tag to describes how the text should be interpreted.

TTS services can struggle when it comes to pronouncing foreign language words and phrases in news articles (quite common!). NewsNet detects foreign language words...

Podcasts have quickly become a modern-day media and are being embraced by publishers as a way to catch their readers on the move.

A recent study published by the Reuters Institute reports almost just as many young people are listening to podcasts as those who use Facebook for news!

SpeechKit makes publishing podcasts simple. Converting text articles into audio allows readers another opportunity to engage with your stories and distributing these through the podcast apps further increases the awarness of your brand.

Setup

If you don’t yet have an account with SpeechKit, go our homepage and set up a free profile. Once you have a project (WordPress, RSS or other) with audio you can start publishing podcasts.

Head to the Distribution tab in the SpeechKit dashboard.

2. From Distribution, select “Add Distribution” and select “Podcast Feed”.

3. Fill in the form to provide info about your podcast project (this will be used in the descriptions in the podcast stores — add a profile image to the project in project settings).

4. You’re now publishing a podcast! If you want to check things out before publishing on Apple Podcast and others, try subscribing to the podcast in iTunes. To do this, copy the RSS feed generated by SpeechKit in the podcast settings (above), then head to iTunes, go to File > Subscribe to Podcast....

Alternatively, there are few good tools to check your feed is working as intended. Here’s a great one from CastFeed . To get your podcast live on Apple Podcast, and other platforms, head to their respective websites and follow the prompts to get your project approved.

More about SpeechKit

At SpeechKit we’re helping 100s of news publishers to automate audio versions of their content efforts. Sign up for a free trial and instantly start creating audio for the next generation of news consumer.