Review of the pronunciation specification and user needs from Abi and EA

Hi APA
Abi James and EA did a review of the pronunciation specification and user
needs on behalf of COGA for the APA responses. The main items are  additional
examples to cover COGA needs and assist with implementation.



Pronunciation User Scenarios
<https://www.w3.org/TR/pronunciation-user-scenarios/>: we would recommend
that an additional scenario is added to cover the challenges with dates,
number, time, Roman numerals, post/zip code and money being misread by text
to speech. We addressed this in one of the design patterns in Content
usable: 4.4.6 Use Clear, Unambiguous Formatting and Punctuation
<https://www.w3.org/TR/coga-usable/#use-clear-unambiguous-formatting-and-punctuation-pattern>.
We would recommend that this scenario is not focussed on education
materials (as there is a maths education one all reader) but could be
related to accessing financial or government services e.g. applying for a
loan or completing a tax return.



Specification for Spoken Presentation In HTML
<https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fraw.githack.com%2Fw3c%2Fpronunciation%2F90425c2688b59b8655ff9c30fbcbb6ebbbfdc72d%2Ftechnical-approach%2Findex.html%23abstract&data=04%7C01%7CA.James%40soton.ac.uk%7C9fa002c9feaa49a992c308d91131423b%7C4a5378f929f44d3ebe89669d03ada9d8%7C0%7C0%7C637559725541146337%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=msyrW9dXnsQ0urLjubbc1jjtY%2FJJrAAEuPEOtY0zo8o%3D&reserved=0>
:

The use of *data-ssml-phoneme-ph *did not include any examples. It was not
clear if a developer would be able to write the phonemes and know they
would be correctly pronounced. Adding examples that can be tested would be
beneficial.



It was not clear from the document how the html <lang> attribute would
interact with this specification alongside an individuals choice of voice
or the voices available on the devices. For example, if the language of the
page is set to en-gb but an individual is using a device where only an
American voice is available then this effects how values of properties such
as date or telephone are announced with the voice default taking precedent
and often misreading (e.g. 4/3/21 is announced as 3rd April 2021 instead of
4th March 2021 as the author intended). For languages where text to speech
may not be available then this is more important that the this mark-up can
be used to override the voice defaults then it can allow the content
creator to write in cultural and linguistic requirements that are not
currently supported by voice engines.



Regards



Abi, EA, Lisa