Skip ⬇︎

Pronunciation Matters

Presenter: Janina Sajka, Mark Hakkinen

Demo by Accessible Platform Architectures Working Group, Pronunciation Task Force

How content is pronounced has long been an accessibility concern, but it's increasingly becoming a mainstream web concern as well with the proliferation of personal assistants and voice interfaces to web content. We propose approaches for controlling pronunciation in spans of web content where uniform (SSML-based) markup can make the difference.

Previous: WCAG Current and Future Work All talks Next: Demo - Web Support for People with Cognitive Disabilities




Hello everyone.

This is Janina Sajka, one of the chairs of Accessible Platform Architectures Working Group at W3C and we have a task force, the Pronunciation Task Force which is creating a specification to allow authors of web content to lock in how text-to-speech engines pronounce.

There is an annoying mispronunciation that screen reader, people have lived with forever, I live at 123 Maple doctor (Dr), (hardly) in the city of Street (St) Paul (hardly).

But it's conceivably also now a mainstream problem because many of us have personal assistant devices that depend on voice interfaces.

So we're here in this brief video to show you some approaches, which could be mainstream, cover our accessibility use cases but also provide authoring the opportunity to help google assistant, to help Siri help Cortana, help Alexa, get it right.

And we have helping us with this demo, Mark Hakkinen from Educational Testing Systems who is kind of intellectually driving a lot of the work in our pronunciation Task Force.

(typing) Alexa, what's the population of Versailles, Indiana?

(beep) As of 2017, the population of Versailles [French city pronunciation], Indiana is 2100 people.

Hmm, I thought that was Versailles.

Alexa told me about Reading, England.

(beep) Here's the Wikipedia article on Reading Berkshire.

Reading is a large historic market town in Berkshire southeastern.

Alexa stop.

Alexa, open pronunciation problems.

Here's your fact Versailles, Indiana is not to be confused with the French city spelled the same way, Versailles.

SSML is supported by a growing number of platforms and allows content authors control over pronunciation and spoken presentation.

Read Aloud capabilities are built into both browsers and available as third party plugins.

Versailles [French city pronunciation], Indiana.

(clicking) Versailles [French city pronunciation], Indiana.

(clicking) Here you're using an experimental editor developed as part of the work of the Pronunciation Task Force to allow us to highlight text in HTML and then add SSML properties to that text - here we're adding pronunciation.

Versailles, Indiana.

Thanks Mark, much appreciated and thanks everyone.

If our approach or something similar is of interest, we are very eager to have your help and expand this to make it useful for the World Wide Web.

Please join us.



Platinum sponsor

Coil Technologies,

Media sponsor


For further details, contact