Unlocking your website's accessibility?

Speech enabling your website

Over the last few years I’ve often been approached by clients and colleagues interested in speech-enabling their websites via a subscription service, wanting to improve accessibility at a stroke.

Speech-enabling means visitors to your site can point and click and have sections of the content read aloud to them. To set this up you just pay a regular subscription to a third party (some systems require installing a plug-in on your website server but not all). Visitors to your site can then point and click at any content and it is read aloud to them (after installing some software on their computer too in most cases).

There are a few of these services out there, such as

Many web masters and communications people think it’s a good idea. It’s an increasingly popular approach (used by the BBC, Twitter, many local authorities, NHS trusts and beyond) and seems like an easy step to take a big leap forward in making your content more accessible.

But does it really make your site more accessible?

Conclusion: nope

The Royal National Institute For The Blind (RNIB), British Deaf Association (BDA) and the Web Accessibility Initiative (WAI) do not recommend use of these kind of site-based speech-enabling services.

They don’t make genuine improvements to the accessibility of the site nor do they help us to meet our accessibility targets; we can only do that by producing accessible code and content.


Limited use of individual technologies

Despite the impressive list of websites which are speech-enabled, the overwhelming majority are not. Out of the billions of websites out there, only a tiny minority use these site-based utilities.

In most cases, if you subscribe to one of these services your visitors will need to install software on their computer before they can use it. Once installed, they are likely to find that the sites which they can use it on are frustratingly limited. Plus, if you install the BrowseAloud software (for example) it only works on subscribed BrowseAloud sites. If a website doesn’t pay a subscription, the software does nothing.

Some of the systems listed above (e.g. ROKTalk, DixerIt and Read Speaker) do not require your visitors to install any software. However, having had a good click around some of the sites which use purely web-based systems, a few of them offered no obvious way to enable the speech and start to use it.

Siemens and Nestle use Read Speaker (according to the Read Speaker site) but I could not see how to access it at all.

NHS Sheffield use ROKTalk which has a clearly visible interface but with an initially off-putting array of buttons. The text resizing and colour changing buttons did not work for me in the way I imagine they were intended, but I only tested them in one browser, feeling perhaps one fail was enough to think there was a problem.

Huntingdonshire District Council use DixerIt which works fairly clearly and simply after clicking a “listen to page” link. Just don’t make the mistake of clicking the link twice otherwise everything crashes and you can no longer navigate the site.

Sadly some of the above point-and-click text-to-speech companies don’t provide links to working examples of their technology on real websites; I couldn’t find any which use Phiware Voice for example, but did notice that they have two followers on Twitter while hunting the web.

Some systems only work on some browsers; for example BrowseAloud doesn’t work in FireFox, ROKTalk doesn’t like iPads and Essential Accessibility is Windows only. This means you may have to change your preferred browser or even device to listen to web content and may have to change again as you move from site to site between systems. This is no solution. In its favour, DixerIt does seem to have good levels of cross-browser / device compatibility.

Clearly these systems do not provide an instant cure-all for your accessibility problems. You still need to prioritise accessibility at every stage of the development process, ensuring that you achieve a good level of standards-compliance, and have content editors with a strong understanding of what makes content accessible and what is best avoided.

Diverts resources

Subscription text-to-speech services cost money which could be spent on genuinely improving the accessibility of the site and thus opening it up to people with:

  • visual impairments,
  • reading difficulties,
  • cognitive impairments,
  • physical disabilities,
  • auditory impairments,
  • non-standard software or hardware,
  • slow connection speeds,
  • and absolutely everyone else too.

Although a prominent logo or interface for a speech-enabled system on your website tells the world that you are aware of the importance of accessibility, it doesn’t necessarily bring about any tangible accessibility benefits which wouldn’t be gained more effectively by focusing on standards compliance.

To make real improvements which benefit everyone we should:

  • ensure our pages meet at least the AA / priority 2 standard for accessibility;
  • simplify / clarify language so it conforms to the Plain English Campaign’s standards for clarity;
  • ensure that content is laid out in the most readable format (good use of white space, bullets, short paragraphs, limited bold text, sub headings, correct use of capitals, all that good stuff);
  • include instructions for visitors to customise the site (change font size, colours, etc.) and point them in the direction of screen reader software / plug-ins which work right across the web;
  • test the site and amend as required.

These steps are sufficient for us to be relatively sure that we have included all the people which speech-enabling systems don’t include, and benefited everyone else as well.

Too reassuring?

Speech enabling services may give web masters with limited experience in accessibility a false sense of security about the accessibility of their site. It’s speech-enabled so we don’t need to worry about accessibility anymore, right? Job done?

Some text-to-speech systems can get a bit carried away with their claims for their software. A claim briefly appeared in BrowseAloud’s promotional material stating it was recommended in the authoritative (but since superseded) “PAS 78 Guide to Good Practice In Commissioning Accessible Websites” when it is NOT recommended in PAS 78.

At the time of writing Phiware Voice claims that “Adding audio to your website with our service will make it accessible to all” although it is hard to understand how it can live up to this considerable boast. It doesn’t make it accessible for people without sound, on a slow connection speed, with physical disabilities, those unable to use a mouse, anyone with significant visual impairments or for the deaf.

Essential Accessibility claim to offer “a software-based service that makes online environments fully accessible to individuals with physical disabilities”. It goes on to say that “Effectively, it ensures that anyone with a physical disability can easily access any website that carries [our logo].” Providing of course that everyone with a physical disability uses Windows to access the web (which they don’t) and are willing and able to install Essential Accessibility’s software (which they may not be).

The claims made by these services are primarily aimed at the site owner, not the visitor. There is a risk that a site owner may sign up to the service and assume that they are now magically fully accessible, overlooking the other steps which they would need to take to achieve reasonable accessibility.

The “why choose us” page on the Phiware Voice website states the following:

Over 20% of the population are “functionally illiterate”. An estimated:

  • 7.5 million people have a reading age of an 11 year old
  • 6 million people are dyslexic
  • 2 million people have a vision impairment
  • 1.5 million have cognitive difficulties
  • 3.4 million have a disability which prevents them from using a standard keyboard, screen and mouse set-up with ease

More than 10% of the population has English as a second language.

These are important points to consider, but not necessarily solved by speech-enabling.  Many of the 7.5 million people who have a reading age of 11 will be deaf and thus will not be helped by speech-enabling. Deafness affects reading ability due to our reliance on our hearing when learning to read; consequently the average reading age among deaf people is 11. For the rest of the population it is 13.

It goes without saying that we need to do more than speech-enable our websites to benefit deaf visitors: we should ensure that our content is written and formatted with great clarity and simplicity. This also benefits the quoted 6 million dyslexia sufferers, 1.5 million people with cognitive difficulties and every single one of the rest of us too.

Point-and-click speech-enablement offers no help with accessibility for people who have difficulties using a keyboard or a mouse, thus mentioning that seems spurious in the above context. Most speech-enablement service are dependent on the visitor being able to see the text and click on it to start the audio reading.

It would be hard for a member of a communications team, keen to improve their website’s accessibility, to see through the hyperbole and grand promises of some of these systems and instead choose to speak to an accessibility expert or take a closer look at their content rather than sign up for a subscription.

What do the Royal National Institute For The Blind (RNIB) have to say about speech enablement?

The E-Access Bulletin of September 2004 considered these text to speech utilities. Three paragraphs in particular are very relevant:

Not everyone is convinced of the benefits to people with impaired vision of speech-enabling the web, however. “Solutions such as these have only limited value to blind and partially sighted people,” says Julie Howell, digital policy development officer at the RNIB. “They are a poor substitute for access solutions such as screen readers and screen magnification.”

Neither does the RNIB approve of the commercial flavour of most speech solutions. “Individual web site owners are required to install the solution on their server or pay to have use of the product on their web site,” she says. “RNIB strongly discourages this approach in favour of the [World Wide Web Consortium’s] Web Accessibility Initiative guidelines. The guidelines are freely available and have the advantage of opening access to all users, using any technology.

“In RNIB’s view, speech-enabled web sites bring nothing to the party and effort would be better spent raising awareness of the WAI guidelines and promoting best practice in web design,” Howell says. “Why should a blind person be restricted only to the sites that have been software-enabled? Surely the great attraction of the web is the endless possibility. The notion that blind and partially sighted people will be satisfied with only the handful of sites that have adopted these solutions must be challenged. RNIB believes that disabled people have the right to visit any web site they choose.”

Beautifully put.

Availability of freeware software which can be used across the web

There are bound to be some people out there who have found services such as BrowseAloud useful on occasions if they find listening easier than reading. I am not discounting the usefulness of simple point-and-click text-to-speech systems for these people; they can help, but are limited and costly forms of help.

There are better alternatives.

Anyone who would benefit from having content read aloud would be better off getting their own screen reader software, thus gaining access to most of those billions of websites instead of just one here and one there. Anyone who has their own screen reader software which they have familiarised themselves with is likely to prefer to use that rather than any of the site-based utilities of varying quality.

  • Freeware screen reader software is available: e.g. Thunder
  • Free-of-charge text-to-speech browser plug-ins: e.g. the Text To Voice Firefox plug-in. This plug-in will convert any text on any site to audio which can either stream instantly or be downloaded as an mp3. Just like any other point-and-click text-to-speech system but it works on any well-constructed site without the site owner having to pay a subscription to anyone.
  • Web-based systems such as WebAnywhere – with a very slight initial learning curve you can use this on any well-constructed site on a reasonable range of browsers and devices.

They have the benefits of not only costing absolutely nothing for both website owner and website users but they also work across the web.

So why do many websites choose to pay a subscription rather than simply encouraging their visitors to try Text To Voice instead free of charge?

I cannot begin to guess. If they can also make genuine attempts to make simple improvements to the accessibility of their content and pledge to keep accessibility as a high priority at every stage of the web development process then they could make a very significant leap forward in running an inclusive website which is easy to use for everybody.

So just who are these subscription services for?

I genuinely have no idea.

If you want to make meaningful improvements in your website’s accessibility, dive into the Web Content Accessibility Guidelines. They are genuinely easy to follow and easy to follow up on. If you need any help, please do get in touch.