FreedomScientific / standards-support

Contains documentation for Vispero software support of Web standards
https://freedomscientific.github.io/standards-support/
GNU General Public License v3.0
110 stars 12 forks source link

Semantic info not always spoken with the default voice #9

Open bart-simons opened 6 years ago

bart-simons commented 6 years ago

Summary

My OS and JAWS are Dutch. Automatic language detection is switched on. Synthesizer is vocalizer expressive. When reading a page with lang="en", its contents is correctly spoken with the English voice. JAWS announces the role of elements such as link, image, heading level. This is nicely done in Dutch and with the default Dutch voice. This is true as long as this announcement preceeds the actual text. But when it is behind, then it is spoken by the English voice that is used to speak the contents.

Example

  1. Go to https://www.gov.uk/
  2. arrow down until you reach the first heading level 1
  3. JAWS reads with the Dutch voice "kopniveau 1" (which is the correct Dutch translation of heading level 1), followed by the actual text of the heading "welcome to gov.uk. The latter is spoken with the English voice. So this is how it should be.
  4. Go back to top with ctrl+home and press h to reach the same heading.
  5. Now you hear "welcome to gov.uk, kopniveau 1". The problem is that both parts are spoken with the English voice. It should switch to the Dutch voice to speak "kopniveau 1".

This seems consistent for semantic info that is spoken after the actual text. For some reason JAWS announces the word link before reading the link text but it reads the value of a button and after that it says that it is a button. Since the link announcement is before, the voice is switched corrrectly. Since the button announcement is at the end, there is no voice swapping and the word knop (translation of button) is spoken with the English voice.

Expected result

JAWS to use the correct voice to speak its own information (such as button, link, heading level, tab selected, checkbox not checked etc). It should not matter whether this info is at the beginning of the string or at the end.

Actual result

Only correct when the JAWS info is read first. When the info follows other text from the web page, JAWS keeps talking with the voice that was used to speak the contents.

Additional Information

JAWS version and build number

JAWS 18.0.4104 (but the issue exists since a long time. It gets more and more annoying with all the new aria announcements)

Operating System and version

Windows 7 64-bit

Browser and version:

IE11 and also in Firefox 56 (but older versions as well)

stevefaulkner commented 6 years ago

thanks @bart-simons will look into both of these related issues.

hanshillen commented 6 years ago

Confirmed,

I have created a test case and have filed this issue (combined with #8) internally as bug 97133: "Dutch JAWS does uses incorrect voice when announcing English content".

dfarough commented 6 years ago

Probably related: I saw this on Webaim list today. A link coded with language of French. the link text is spoken with the French voice as expected, but the title text is spoken using the English voice. Example code was: 2

bart-simons commented 6 years ago

@dfarough yes this is a combination of both issues. It reads 2 with the French voice, then "bezocht link" (Dutch translation of visited link) with the French voice (instead of Dutch), and then the title text with the Dutch voice (instead of French).

stevefaulkner commented 1 year ago

@hanshillen can you check with latest JAWS to see if this is still an issue?

hanshillen commented 1 year ago

I did some testing with JAWS 2023.2307.37. The issue is still occurring: in the scenario described above, JAWS announces "kopniveau X" with the Dutch voice when it's prefixed to the heading text (e.g. when using arrow keys navigation), and with the English voice when it comes after the heading text (e.g. when using heading navigation).

BrettLewisVispero commented 1 year ago

Imported into ADO as bug 40481