Tracking.

The default should remain change the pitch as we are doing now.  The alternative would be to speak the word "cap" before the upper-case letter.  

Add accessibility keyword.  Apologies for spam.

In Vietnamese, you would say « Chữ hoa » before capital letters, and « Chữ thường » before ordinary letters. I suppose that is like "Upper-case" and "Lower-case".

This is important, because meaning in our language depends on tone and pitch. Changing the pitch of a character changes the character, and changing the pitch of a word will actually change its _meaning_. So changing the pitch to convey upper-case will seriously confuse a Vietnamese user.

We have 72 vowels, for example, differentiated by tone and pitch, which is indicated by combined diacritics.

I hope this information is useful. I can provide more details when you want them.

Clytie Siddall
Vietnamese Free-Software Translation Team

Thanks Clytie!  Tonal languages are definitely something to be aware of when considering use of pitch to convey other meanings, so thanks for the very useful reminder.  :-)

The default voice selection for upper case happens to be one that uses a higher pitch baseline right now because that is what is expected by the vast majority of our users.  The good thing is that the voice selection is completely customizable by the user, including the use (or not) of a higher pitch baseline for uppercase indication.  The even better thing may be that verbalized capitalization might be a good solution to help with the problems we'll face with tonal languages.

Thanks for all your work on translating!

The UI for this should be as follows: 
On the speech tab we should add a "capital letters" panel.  This panel should contain two radio buttons.  They will be: "raise pitch" and "say cap".  Raise pitch should be the default.  

Just a note: some synthesizers support this directly.  For example, espeak does this: http://espeak.sourceforge.net/commands.html.

See also bug 523012.

I'm putting this an some other speech-related bugs on our 2.28 list.  My thinking is that we need to try to push these things as close to the speech engine as possible.  With the thought of moving from gnome-speech to something else, we need to make sure that "something else" can support this kind of thing.

We are late in the 2.28 release cycle and I want to focus on "high impact"/"low risk" items that also fall within the release team's restrictions in place.  Regretfully, this bug doesn't fit well within those constraints and we'll review it for the 2.29 release cycle.

It's possible to use the speechd.client.SSIPClient.set_cap_let_recogn() methode to set the capital letter recognizing. This can be set to none, spell and icon. For 'icon' the sound-icons are required.

I think the best solution would be to use the 'icon'. Because, depending on the language 'spell' could be too much. The sound-icon isn't very intrusive.