search

w3c / pronunciation

Pronunciation Task Force deliverables
https://www.w3.org/WAI/APA/task-forces/pronunciation/
Other
20 stars 12 forks source link

Compatibility with platform accessibility APIs? #88

Open minorninth opened 3 years ago

minorninth commented 3 years ago

Do we have any input from (1) the owners of accessibility APIs, such as IAccessible2, UI Automation, ATK, NSAccessibility, Android, etc. on how this information could be conveyed to AT by the user agent?

I believe that both Android and macOS already have technical ways to specify pronunciation rules in text that's returned from a user agent via accessibility APIs. I think this specification wouldn't be complete until it contained a complete and comprehensive implementation guide for how to transform the specified SSML into those platform-specific representations.

alia11y commented 3 years ago

Thank you for your comments, Dominic. we are seeking inputs from the owner of a11y API.

leo60228 commented 1 year ago

As a disclaimer, I don't have any background in platform accessibility APIs, and I'm looking at this spec from the perspective of a web developer trying to ensure correct pronunciation from screen readers. However, as far as I can tell, only UIAccessibility on iOS has existing support for pronunciation, though SwiftUI's speechPhoneticRepresentation method is listed as supported on macOS 12.0+.

On iOS, there are five speech attributes for text as part of UIAccessibility. Of these, only accessibilitySpeechIPANotation maps cleanly to the spec (data-ssml-phoneme-ph). accessibilitySpeechLanguage specifies the language of the content, which is explicitly the lang attribute's job. accessibilitySpeechPitch is conceptually the same as data-ssml-prosody-pitch, but the former is a 0.0-1.0 scale and it's not clear how the latter should be mapped to this. accessibilitySpeechPunctuation is a boolean for whether punctuation should be spoken. The SSML spec implies that this would map to data-ssml-say-as-detail, but the note on say-as values doesn't back this up. The final attribute is accessibilitySpeechQueueAnnouncement, which from my understanding is only relevant to notifications.

leo60228 commented 1 year ago

I think this specification wouldn't be complete until it contained a complete and comprehensive implementation guide for how to transform the specified SSML into those platform-specific representations.

Emphasizing that I don't have background experience with platform accessibility APIs, but I feel like this should be specified by them. I'm not sure how easy it'd be to get in touch, but maybe at least the GNOME people could be interested?