Using the NVDA Dictionaries to Alter Pronunciations


Robert Doc Wright godfearer
 

any time I have the folowing:
02 Mark
02_Mark
NVDA pronounces it MarchTwonk.
What can I do to get it to say 02 Mark? 
*********
family Times weekly Movie times
 

go to Doc's Stream
http://stream.wrighthere.net:8000/stream.mp3
Ask Miss A to play family times on tuneIn
You can also find family times on OoTunes.
 
 

walking in VictoryI welcome all that are interested in joining our devotionals on Monday thru saturdays at 8am EST. join the devotion channel.
Also, there is a weekly bible study at 8pm EST in the Walking with Jesus channel.
 

Add the following information in your TeamTalk
host address: wiv.wrighthere.net
TCP/UDP ports: 10333
 
entry name: walking in Victory
tab twice and press space on Add/update
 


 

1. Open NVDA Default Dictionary, NVDA+N,P,D,D
 
2. Activate Add Button, ALT+A
 
3. In Add Dictionary Entry Dialog, put the following in the Pattern edit box:  \s0?2[ _]Mark\s
    I am presuming the actual text will have white space before and after it.  If not, we'll have to tweak the pattern.
 
4. In the Replacement edit box, enter what you want the screen reader to say.  In this case I'd presume
    that would be either "zero two Mark" or "oh two Mark".  The quotation marks are NOT needed.
 
5. If you want a note for yourself as to what this dictionary entry is for, put that in the comment 
    edit box.  I find it's a good idea to do this when it's not really clear what a regular expression may do
    or the replacement doesn't make that clear.
 
6. In the Type group, activate the radio button for Regular expression.
 
7. Activate the OK button to save.  You should now hear whatever you put in the Replacement every time
    the Pattern is matched in any text that NVDA is reviewing.
--

Brian - Windows 10 Pro, 64-Bit, Version 20H2, Build 19042  

[Regarding the Supreme Court refusing to hear the case brought by Texas to overturn the votes certified by 4 states:Pleased with the SCOTUS ruling, but also immediately slightly terrified of where this crazy train goes next.  We should know by now there’s a bottomless supply of crazy.

        ~ Brendan Buck, former adviser to Speakers of the House Paul Ryan and John Boehner 

 


 

Downloadable set of step-by-step instructions:  Using the NVDA Dictionaries with Regular Expression Matching to Change Pronunciations (docx)

 


I also added a bit of discussion about both the Whole Word and Anywhere Type options.  I'll also add my warning about the Anywhere type again, here:  If it's to be used at all, it must be used with extreme caution.  I strongly suggest it never be used.  See the tutorial for an example of why.
--

Brian - Windows 10 Pro, 64-Bit, Version 20H2, Build 19042  

[Regarding the Supreme Court refusing to hear the case brought by Texas to overturn the votes certified by 4 states:Pleased with the SCOTUS ruling, but also immediately slightly terrified of where this crazy train goes next.  We should know by now there’s a bottomless supply of crazy.

        ~ Brendan Buck, former adviser to Speakers of the House Paul Ryan and John Boehner 

 


 

Robert,

           Please use the pattern string that is in the downloadable instructions:  \s?0?2[ _]Mark\s

           I changed it such that if what you want pronounced as, "oh two Mark," were to occur as the very first part of the any line, it would still match.  A single question mark character makes all the difference between not matching and matching at the start of a line.

--

Brian - Windows 10 Pro, 64-Bit, Version 20H2, Build 19042  

[Regarding the Supreme Court refusing to hear the case brought by Texas to overturn the votes certified by 4 states:Pleased with the SCOTUS ruling, but also immediately slightly terrified of where this crazy train goes next.  We should know by now there’s a bottomless supply of crazy.

        ~ Brendan Buck, former adviser to Speakers of the House Paul Ryan and John Boehner 

 


Quentin Christensen
 

Do you have any speech dictionary entries?

For me, using eSpeak-NG, OneCore Voices or SAPI 5, I get "Zero Two Mark" for both, and for the second, at higher punctuation level I get Zero Two Line Mark".

On Thu, Dec 17, 2020 at 9:46 PM Robert Doc Wright godfearer <godfearer@...> wrote:
any time I have the folowing:
02 Mark
02_Mark
NVDA pronounces it MarchTwonk.
What can I do to get it to say 02 Mark? 
*********
family Times weekly Movie times
 

go to Doc's Stream
http://stream.wrighthere.net:8000/stream.mp3
Ask Miss A to play family times on tuneIn
You can also find family times on OoTunes.
 
 

walking in VictoryI welcome all that are interested in joining our devotionals on Monday thru saturdays at 8am EST. join the devotion channel.
Also, there is a weekly bible study at 8pm EST in the Walking with Jesus channel.
 

Add the following information in your TeamTalk
host address: wiv.wrighthere.net
TCP/UDP ports: 10333
 
entry name: walking in Victory
tab twice and press space on Add/update
 



--
Quentin Christensen
Training and Support Manager


Jeff Samco
 

When I use Code Factory's Eloquence I get the same result as Robert. When I switch to a OneCore voice I get the correct pronunciation. I think Eloquence has some built-in substitutions That takes the Mar of Mark and reads it as the month of March and tries to render the entire string as a date. I have noticed the 3-letter string of Dec, when occuring as the first part of a word, can be spoken as December. I have checked my extensive pronunciation dictionary and find nothing that might be triggering this from that aspect.

Jeff


At 03:59 PM 12/17/2020, you wrote:
Do you have any speech dictionary entries?

For me, using eSpeak-NG, OneCore Voices or SAPI 5, I get "Zero Two Mark" for both, and for the second, at higher punctuation level I get Zero Two Line Mark".

On Thu, Dec 17, 2020 at 9:46 PM Robert Doc Wright godfearer <godfearer@...> wrote:
any time I have the folowing:
02 Mark
02_Mark
NVDA pronounces it MarchTwonk.
What can I do to get it to say 02 Mark?Â
*********
family Times weekly Movie times
Â

go to Doc's Stream
http://stream.wrighthere.net:8000/stream.mp3
Ask Miss A to play family times on tuneIn
You can also find family times on OoTunes.
Â
Get the stream guide:
http://www.wrighthere.net/FamilyTimes/playingnow.html
********
Â

walking in VictoryI welcome all that are interested in joining our devotionals on Monday thru saturdays at 8am EST. join the devotion channel.
Also, there is a weekly bible study at 8pm EST in the Walking with Jesus channel.
Â

Add the following information in your TeamTalk
host address: wiv.wrighthere.net
TCP/UDP ports: 10333
Â
entry name: walking in Victory
tab twice and press space on Add/update
Â
or
https://www.dropbox.com/s/zfa8s6ragw0yltb/WalkingInVictory.tt?dl=1



--
Quentin Christensen
Training and Support Manager

Web:Â www.nvaccess.orgÂ
Training:Â https://www.nvaccess.org/shop/
Certification:Â https://certification.nvaccess.org/
User group:Â https://nvda.groups.io/g/nvda
Facebook:Â http://www.facebook.com/NVAccessÂ
Twitter: @NVAccess


 

I am making the presumption here that the reference is to the Biblical Gospel of Mark, and, if so, am wondering if the issue exhibits itself only for Chapter 2, or for other or all chapters when the format shown for the chapter two reference is used for any of the other chapters?

If it's a "cross chapter" issue, then the solution I gave would need to be tweaked again, so that any chapter number, with or without a leading zero, would be captured correctly and the replacement string would have the correct chapter supplied rather than relying on "oh two" alone.

If we're talking about all chapters for all four Gospels, I think this regular expression would work for the pattern:  \b0?([1-9]+)[ _](Matthew|Mark|Luke|John)\b
where the replacement would be:  \1 \2   
which could still be a problem depending on the synth and voice
or, potentially: \2 Chapter \1
--

Brian - Windows 10 Pro, 64-Bit, Version 20H2, Build 19042  

[Regarding the Supreme Court refusing to hear the case brought by Texas to overturn the votes certified by 4 states:Pleased with the SCOTUS ruling, but also immediately slightly terrified of where this crazy train goes next.  We should know by now there’s a bottomless supply of crazy.

        ~ Brendan Buck, former adviser to Speakers of the House Paul Ryan and John Boehner 

 


Robert Doc Wright godfearer
 


Ok then I need to address this to Code factory because I am using Eloquence.

----- Original Message -----
Sent: Thursday, December 17, 2020 4:59 PM
Subject: Re: [nvda] pronunciations

Do you have any speech dictionary entries?

For me, using eSpeak-NG, OneCore Voices or SAPI 5, I get "Zero Two Mark" for both, and for the second, at higher punctuation level I get Zero Two Line Mark".

On Thu, Dec 17, 2020 at 9:46 PM Robert Doc Wright godfearer <godfearer@...> wrote:
any time I have the folowing:
02 Mark
02_Mark
NVDA pronounces it MarchTwonk.
What can I do to get it to say 02 Mark? 
*********
family Times weekly Movie times
 

go to Doc's Stream
http://stream.wrighthere.net:8000/stream.mp3
Ask Miss A to play family times on tuneIn
You can also find family times on OoTunes.
 
 

walking in VictoryI welcome all that are interested in joining our devotionals on Monday thru saturdays at 8am EST. join the devotion channel.
Also, there is a weekly bible study at 8pm EST in the Walking with Jesus channel.
 

Add the following information in your TeamTalk
host address: wiv.wrighthere.net
TCP/UDP ports: 10333
 
entry name: walking in Victory
tab twice and press space on Add/update
 



--
Quentin Christensen
Training and Support Manager


 

On Fri, Dec 18, 2020 at 04:07 AM, Robert Doc Wright godfearer wrote:
Ok then I need to address this to Code factory because I am using Eloquence.
-
No snark toward you intended (it's aimed at Code Factory/Eloquence):  Good luck with that.

This sort of behavior is not limited to Eloquence, and I've experienced it with a number of other synths, and when using them with other screen readers.  The issue with making changes is the same for a synth-maker as it is for any software maker that has had a given behavior or feature out there for a very long time:  Many users rely on that behavior, even if it drives other users insane.  You cannot please everyone with whatever defaults or "efficiency shortcuts" you take that work for many, but really don't work for some (who may still be many in number).

The whole purpose of the speech dictionary functions in screen readers is to get around quirks like this one.  And if you want a quick result, you use that tool yourself, as it could be months to years to never as far as getting any change from the synth itself.
--

Brian - Windows 10 Pro, 64-Bit, Version 20H2, Build 19042  

[Regarding the Supreme Court refusing to hear the case brought by Texas to overturn the votes certified by 4 states:Pleased with the SCOTUS ruling, but also immediately slightly terrified of where this crazy train goes next.  We should know by now there’s a bottomless supply of crazy.

        ~ Brendan Buck, former adviser to Speakers of the House Paul Ryan and John Boehner 

 


Dan Miner
 

I have been getting  similar issues with versions of source code packages.  I’m using MS Core for my voice but it is clearly interpreting things like.  20.01.05 as a date... my favorite is partial intepretation..  2.03.26 comes out as 1 March 2016.  *sigh*

    Dan

On Dec 18, 2020, at 9:54 AM, Brian Vogel <britechguy@...> wrote:

On Fri, Dec 18, 2020 at 04:07 AM, Robert Doc Wright godfearer wrote:
Ok then I need to address this to Code factory because I am using Eloquence.
-
No snark toward you intended (it's aimed at Code Factory/Eloquence):  Good luck with that.

This sort of behavior is not limited to Eloquence, and I've experienced it with a number of other synths, and when using them with other screen readers.  The issue with making changes is the same for a synth-maker as it is for any software maker that has had a given behavior or feature out there for a very long time:  Many users rely on that behavior, even if it drives other users insane.  You cannot please everyone with whatever defaults or "efficiency shortcuts" you take that work for many, but really don't work for some (who may still be many in number).

The whole purpose of the speech dictionary functions in screen readers is to get around quirks like this one.  And if you want a quick result, you use that tool yourself, as it could be months to years to never as far as getting any change from the synth itself.
--

Brian - Windows 10 Pro, 64-Bit, Version 20H2, Build 19042  

[Regarding the Supreme Court refusing to hear the case brought by Texas to overturn the votes certified by 4 states:Pleased with the SCOTUS ruling, but also immediately slightly terrified of where this crazy train goes next.  We should know by now there’s a bottomless supply of crazy.

        ~ Brendan Buck, former adviser to Speakers of the House Paul Ryan and John Boehner