alphanumeric strings as phonetics?


Chris Smart
 

Would anyone else like to see NVDA be able to pronounce alphanumeric strings using the phonetic alphabet?

This would be useful for Canadian and British postal codes, ham radio call signs, and probably other things as well.

 

L9G3S6 would be Lima 9 Golf 3 Siera 6.

 

Is that something I should post to Github about?

 

Jaws can do this by the way.

 


 

Isn't this already handled by the following add-on?:

Enhanced Phonetic Reading (EnhancedPhoneticReading)
add-on

The development channel has been tested with NVDA 2022 and it's rather certain that this will eventually become the production release.  The stable version is tested up through 2021.3
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


Chris Smart
 

Not exactly, or at least not how I’m able to read things with the other screen reader.

With this add-on, I move character by character, and hear the phonetics for each character after I hear the letter name, and after an adjustable pause.

 

In Jaws, if I turn on “Spell Alphanumeric Data”, I have the following options:

Off, which is the default setting, reads the text as it appears. For example, the license plate number 123RON is read as one hundred twenty-three Ron. The Spell option spells alphanumeric data. For example, 123RON is read as 1 2 3 R O N. Spell Phonetically uses the phonetic alphabet to spell alphanumeric data. For example, JAWS reads 123RON as 1 2 3 Romeo Oscar November.

 

It’s that third behavior which I would love to have in NVDA.

 

Chris

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian Vogel
Sent: July 2, 2022 11:57 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] alphanumeric strings as phonetics?

 

Isn't this already handled by the following add-on?:

Enhanced Phonetic Reading (EnhancedPhoneticReading)
add-on

The development channel has been tested with NVDA 2022 and it's rather certain that this will eventually become the production release.  The stable version is tested up through 2021.3
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


 

Chris,

The way I read the following in the documentation, "Read the character description instead of the current character. This feature must be enabled or disabled manually and will be deactivated when NVDA is closed," suggested to me that what you wanted is able to be turned on, but only on a "per NVDA session" basis.

Have you played with that option?
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


Chris Smart
 

Hi Brian.

I couldn’t find that option when I went into NVDA> Preferences > Settings > Enhanced Phonetic Reading.

I see two options: A checkbox that says “enable delayed descriptions for characters”, and a slider to adjust how many milliseconds gap there is between the letter and the phonetic.

Maybe I’m missing something somewhere else though.

 

I’m talking about the ability to hear a string phonetically.

 

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian Vogel
Sent: July 2, 2022 12:17 PM
To: nvda@nvda.groups.io
Subject: Re: [nvda] alphanumeric strings as phonetics?

 

Chris,

The way I read the following in the documentation, "Read the character description instead of the current character. This feature must be enabled or disabled manually and will be deactivated when NVDA is closed," suggested to me that what you wanted is able to be turned on, but only on a "per NVDA session" basis.

Have you played with that option?
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


Sarah k Alawami
 

I know what he means. The ham calls sign  or something, similar to the settings on the bNS. I personally would not go for it, why not create a poll on straw poll and see where that gets you. Then you can decide if you want to create a suggestion on github.

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Chris Smart
Sent: Saturday, July 2, 2022 9:29 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] alphanumeric strings as phonetics?

 

Hi Brian.

I couldn’t find that option when I went into NVDA> Preferences > Settings > Enhanced Phonetic Reading.

I see two options: A checkbox that says “enable delayed descriptions for characters”, and a slider to adjust how many milliseconds gap there is between the letter and the phonetic.

Maybe I’m missing something somewhere else though.


 

On Sat, Jul 2, 2022 at 12:39 PM, Sarah k Alawami wrote:
I know what he means.
-
As do I.

Here's more from the documentation if you use the Add-On Help feature in Add-Ons Manager:
----

Usage

The add-on functionality will be enabled once you install it.
To enable or disable it, go to NVDA settings and select "Enhanced phonetic reading". In that category you can set the following parameters:

  • Enable delayed descriptions for characters.
  • Delay to announce character descriptions (in ms): the time the add-on waits to speak the description for the read character. You can't set it greater than 20000 ms.

Scripts

  • toggle instant character descriptions: Assigned to "nvda + control + numpad2" or "nvda+ control + enter" for laptop keyboards. This script enables or disables the instant character descriptions feature.
  • toggle delayed descriptions: no gestures assigned. This script lets you toggle the delayed descriptions status.

You can assign a command or gesture in speech category in the NVDA input gestures dialog.

----
 
It appears to me that you can get instant phonetic reading, and only phonetic reading, either by unchecking the checkbox for delayed descriptions or using the toggle keyboard command to turn it off.

I understand why many would not want to hear the letter first followed by its phonetic equivalent.  I'm actually surprised that this add-on is not set up with defaults that are the reverse of what they are, but it's configurable.

I used the dev version since I do have NVDA 2022 installed.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


Gene
 

I'm curious about the JAWS feature.  Can you set the speed at which letters are announced and whether and how long ;pauses are between the announcement of individual letters?  I would think that if that weren't done, the letters might be announced so quickly you would lose track of them if you use a somewhat fast speech rate.

Gene

On 7/2/2022 11:39 AM, Sarah k Alawami wrote:

I know what he means. The ham calls sign  or something, similar to the settings on the bNS. I personally would not go for it, why not create a poll on straw poll and see where that gets you. Then you can decide if you want to create a suggestion on github.

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Chris Smart
Sent: Saturday, July 2, 2022 9:29 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] alphanumeric strings as phonetics?

 

Hi Brian.

I couldn’t find that option when I went into NVDA> Preferences > Settings > Enhanced Phonetic Reading.

I see two options: A checkbox that says “enable delayed descriptions for characters”, and a slider to adjust how many milliseconds gap there is between the letter and the phonetic.

Maybe I’m missing something somewhere else though.



Chris Smart
 

I don’t find it hard to understand, and I listen to Eloquence at about 70%.

With the feature turned on and set to phonetics, it says each word as if there is a comma after it. So, for example, L9G3S6 will be read as “Lima, 9, golf, 3, Sierra, 6”.

 


Gene
 

That's what I had in mind, the pause would be important, I think.

Gene

On 7/2/2022 12:14 PM, Chris Smart wrote:

I don’t find it hard to understand, and I listen to Eloquence at about 70%.

With the feature turned on and set to phonetics, it says each word as if there is a comma after it. So, for example, L9G3S6 will be read as “Lima, 9, golf, 3, Sierra, 6”.

 



 

On Sat, Jul 2, 2022 at 01:15 PM, Chris Smart wrote:
With the feature turned on and set to phonetics, it says each word as if there is a comma after it. So, for example, L9G3S6 will be read as “Lima, 9, golf, 3, Sierra, 6”.
-
So, if I'm understanding you correctly, you want NVDA to detect that it has an unpronounceable alphanumeric string when it's doing a say all and, when such is encountered, say it phonetically?

It's not that something like this couldn't be done, but it does require quite a bit of overhead as each "word" would need to be analyzed as to whether it was an alphanumeric string in realtime, which is a non-trivial and resource intensive task (relatively speaking) since there is no way to predict where such might occur in a text.  Does the JAWS feature do this during a Say All?

And I did just play with the Enhanced Phonetic Reading add-on and if I use the NVDA + CTRL + NumPad 2 command I can toggle instant reading of each character phonetically and have it skip the regular announcement of the letter itself.  So, provided I'm going letter by letter, I get “Lima, 9, golf, 3, Sierra, 6” for L9G3S6. 
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


Chris Smart
 

So, if I'm understanding you correctly, you want NVDA to detect that it has an unpronounceable alphanumeric string when it's doing a say all and, when such is encountered, say it phonetically?

Yes. That would be ideal.


It's not that something like this couldn't be done, but it does require quite a bit of overhead as each "word" would need to be analyzed as to whether it was an alphanumeric string in realtime, which is a non-trivial and resource intensive task (relatively speaking) since there is no way to predict where such might occur in a text.  Does the JAWS feature do this during a Say All?

Yep, I just tried it.

 


Sarah k Alawami
 

Yep I listen to nvda at close to 100, or at least I’m retraining my ear. And for me I don’t really find alpha bravo etc hard.

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Chris Smart
Sent: Saturday, July 2, 2022 10:15 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] alphanumeric strings as phonetics?

 

I don’t find it hard to understand, and I listen to Eloquence at about 70%.

With the feature turned on and set to phonetics, it says each word as if there is a comma after it. So, for example, L9G3S6 will be read as “Lima, 9, golf, 3, Sierra, 6”.

 


Ron Canazzi
 

Hi Group,

A simple way of describing this is indeed to have 'spell alpha-numeric expressions phonetically' active during screen reader 'say all' reading.  Here's an example. 

Say I have the following sentence in 'say all.' mode.
I just had a conversation with Robert smith WA2PDQ on the forty meter ham band. 
With 'spell alpha-numeric expression phonetically' turned on in JAWS, this sentence would be rendered in 'say all' mode as follows:
I just had a conversation with Robert smith Whiskey Alpha 2 Papa Delta Quebec on the forty meter ham band. 

This would be a nice feature if it could be implemented in NVDA.

On 7/2/2022 2:41 PM, Brian Vogel wrote:
On Sat, Jul 2, 2022 at 01:15 PM, Chris Smart wrote:
With the feature turned on and set to phonetics, it says each word as if there is a comma after it. So, for example, L9G3S6 will be read as “Lima, 9, golf, 3, Sierra, 6”.
-
So, if I'm understanding you correctly, you want NVDA to detect that it has an unpronounceable alphanumeric string when it's doing a say all and, when such is encountered, say it phonetically?

It's not that something like this couldn't be done, but it does require quite a bit of overhead as each "word" would need to be analyzed as to whether it was an alphanumeric string in realtime, which is a non-trivial and resource intensive task (relatively speaking) since there is no way to predict where such might occur in a text.  Does the JAWS feature do this during a Say All?

And I did just play with the Enhanced Phonetic Reading add-on and if I use the NVDA + CTRL + NumPad 2 command I can toggle instant reading of each character phonetically and have it skip the regular announcement of the letter itself.  So, provided I'm going letter by letter, I get “Lima, 9, golf, 3, Sierra, 6” for L9G3S6. 
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

The real art of conversation is not only to say the right thing in the right place but to leave unsaid the wrong thing at the tempting moment.

        ~ Dorothy Nevill


-- 
Signature:
For a nation to admit it has done grievous wrongs and will strive to correct them for the betterment of all is no vice;
For a nation to claim it has always been great, needs no improvement  and to cling to its past achievements is no virtue!


Brian's Mail list account
 

Could it just be looking at words with caps inside the word other than at the start?

I'd still expect a lag though, unless it had already built up an index of such things in the document being read first which would delay the start of reading.
Goodness knows what it would make of nvda Python logs and code then! Grin

Brian

--
bglists@...
Sent via blueyonder.(Virgin media)
Please address personal E-mail to:-
briang1@..., putting 'Brian Gaff'
in the display name field.

----- Original Message -----
From: "Chris Smart" <ve3rwj@...>
To: <nvda@nvda.groups.io>
Sent: Saturday, July 02, 2022 7:48 PM
Subject: Re: [nvda] alphanumeric strings as phonetics?


So, if I'm understanding you correctly, you want NVDA to detect that it has an unpronounceable alphanumeric string when it's doing a say all and, when such is encountered, say it phonetically?



Yes. That would be ideal.


It's not that something like this couldn't be done, but it does require quite a bit of overhead as each "word" would need to be analyzed as to whether it was an alphanumeric string in realtime, which is a non-trivial and resource intensive task (relatively speaking) since there is no way to predict where such might occur in a text. Does the JAWS feature do this during a Say All?



Yep, I just tried it.


Chris Smart
 

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian's Mail list account via groups.io
Sent: July 3, 2022 4:42 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] alphanumeric strings as phonetics?

Could it just be looking at words with caps inside the word other than at the start?

No.

I type ve3rwj and hear "victor, echo, 3, Romeo, whiskey, Juliet".
And there is no pause between that string and others in a sentence, so it can't be too computationally demanding.

I just experimented a bit more, and strings starting with up to three letters are read phonetically, such as abc3 (alpha, bravo, Charlie, 3). But as soon as I add a fourth letter, Jaws goes back to trying to pronounce it as a word, for example abcd4.