Nvda and the International phonetic Alphabet, (IPA)


Leo
 


 

Hello,

I have noticed that  espeak reads some symbols of the international phonetic alphabet; however, this can not be done by other synthetic voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of the synthetizer?

Or, how to make that other synthetizers read those symbols?  For instance, vocalizer.

 

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

 

So, is there a file inside the installation folder of nvda or user settings  where I can paste  that syntaxis to have nvda reading all those symbols?

 

Thanks for any info.

 

--
Literophilus


Teri McElroy <shimzin.lists@...>
 

Hello Lev

I have just recently emailed the list to find out about this same topic. You can go to the nvda menu, then preferences, punctuation/symbol pronunciation. You'll need to input the symbols along with how you want the speech engine to pronounce them. However, I don't think putting in the unicode values is going to work. You'll need to input the actual symbol along with how you want NVDA to pronounce it. I've just tried entering a unicode value here and it didn't work for me. But when I enter the actual symbol, it does.

You can find information about punctuation/symbol pronunciation in section 11.2.2 of the online userguide. It will explain about inputting symbols.

Hope that helps.


Teri

On 13-Mar-19 8:18 AM, Lev wrote:


Hello,

I have noticed that espeak reads some symbols of the international phonetic alphabet; however, this can not be done by other synthetic voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of the synthetizer?

Or, how to make that other synthetizers read those symbols?For instance, vocalizer.

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

So, is there a file inside the installation folder of nvda or user settings where I can paste that syntaxis to have nvda reading all those symbols?

Thanks for any info.

--
Literophilus


Brian's Mail list account <bglists@...>
 

Yes this seems to be a shortcoming of the synths themselves to my way of thinking and as such really out of the control of nvda. Everyone seems to make their synths work slightly differently and I guess unless somebody has solved this and has the right text files to make them sound right, you have to do it the hard way.
Brian

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

----- Original Message -----
From: "Teri McElroy" <shimzin.lists@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Tuesday, March 12, 2019 8:21 PM
Subject: Re: [nvda] Nvda and the International phonetic Alphabet, (IPA)


Hello Lev

I have just recently emailed the list to find out about this same topic. You can go to the nvda menu, then preferences, punctuation/symbol pronunciation. You'll need to input the symbols along with how you want the speech engine to pronounce them. However, I don't think putting in the unicode values is going to work. You'll need to input the actual symbol along with how you want NVDA to pronounce it. I've just tried entering a unicode value here and it didn't work for me. But when I enter the actual symbol, it does.

You can find information about punctuation/symbol pronunciation in section 11.2.2 of the online userguide. It will explain about inputting symbols.

Hope that helps.


Teri


On 13-Mar-19 8:18 AM, Lev wrote:


Hello,

I have noticed that espeak reads some symbols of the international phonetic alphabet; however, this can not be done by other synthetic voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of the synthetizer?

Or, how to make that other synthetizers read those symbols?For instance, vocalizer.

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

So, is there a file inside the installation folder of nvda or user settings where I can paste that syntaxis to have nvda reading all those symbols?

Thanks for any info.

--
Literophilus


Leo
 

thank you Teri,
however, there are many many symbols that I want to include and adding
one by one is something that I don't want to do...  I prefer just
continue using the other screen reader for this purpose...
From what I understood, you need to read some symbols from the IPA, so
if you are interested and you are already a user of the other screen
reader, inbox me and I tell you how I did it.


Literophilus

El 12/03/2019 a las 14:21, Teri McElroy escribió:
Hello Lev

I have just recently emailed the list to find out about this same
topic. You can go to the nvda menu, then preferences,
punctuation/symbol pronunciation. You'll need to input the symbols
along with how you want the speech engine to pronounce them. However,
I don't think putting in the unicode values is going to work. You'll
need to input the actual symbol along with how you want NVDA to
pronounce it. I've just tried entering a unicode value here and it
didn't work for me. But when I enter the actual symbol, it does.

You can find information about punctuation/symbol pronunciation in
section 11.2.2 of the online userguide. It will explain about
inputting symbols.

Hope that helps.


Teri


On 13-Mar-19 8:18 AM, Lev wrote:


Hello,

I have noticed that espeak reads some symbols of the international
phonetic alphabet; however, this can not be done by other synthetic
voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of the
synthetizer?

Or, how to make that other synthetizers read those symbols?For
instance, vocalizer.

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

So, is there a file inside the installation folder of nvda or user
settings where I can paste that syntaxis to have nvda reading all
those symbols?

Thanks for any info.

--
Literophilus


Leo
 

well if the situation is such as nobody can answer how to do this in
nvda, then I must agree with you,
there should be a text file wher where you can easily follow simple
syntaxis and paste whatever symbols you want
without having to add one by one.

Literophilus

El 13/03/2019 a las 2:35, Brian's Mail list account via Groups.Io escribió:
Yes this seems to be a shortcoming of the synths themselves to my way
of thinking and as such really out of the control of nvda. Everyone
seems to make  their synths work slightly differently and  I guess
unless somebody has solved this and has the right text files to make
them sound right, you have to do it the hard way.
Brian

bglists@blueyonder.co.uk
Sent via blueyonder.
Please address personal E-mail to:-
briang1@blueyonder.co.uk, putting 'Brian Gaff'
in the display name field.
----- Original Message ----- From: "Teri McElroy"
<shimzin.lists@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Tuesday, March 12, 2019 8:21 PM
Subject: Re: [nvda] Nvda and the International phonetic Alphabet, (IPA)


Hello Lev

I have just recently emailed the list to find out about this same
topic. You can go to the nvda menu, then preferences,
punctuation/symbol pronunciation. You'll need to input the symbols
along with how you want the speech engine to pronounce them. However,
I don't think putting in the unicode values is going to work. You'll
need to input the actual symbol along with how you want NVDA to
pronounce it. I've just tried entering a unicode value here and it
didn't work for me. But when I enter the actual symbol, it does.

You can find information about punctuation/symbol pronunciation in
section 11.2.2 of the online userguide. It will explain about
inputting symbols.

Hope that helps.


Teri


On 13-Mar-19 8:18 AM, Lev wrote:


Hello,

I have noticed that espeak reads some symbols of the international
phonetic alphabet; however, this can not be done by other synthetic
voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of
the synthetizer?

Or, how to make that other synthetizers read those symbols?For
instance, vocalizer.

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

So, is there a file inside the installation folder of nvda or user
settings where I can paste that syntaxis to have nvda reading all
those symbols?

Thanks for any info.

--
Literophilus



Teri McElroy <shimzin.lists@...>
 

Hi Lev

Thanks, I already have "the other screen reader" working with IPA but need NVDA to do it too.

On 14-Mar-19 1:48 AM, Lev wrote:
thank you Teri,
however, there are many many symbols that I want to include and adding
one by one is something that I don't want to do...  I prefer just
continue using the other screen reader for this purpose...
From what I understood, you need to read some symbols from the IPA, so
if you are interested and you are already a user of the other screen
reader, inbox me and I tell you how I did it.


Literophilus

El 12/03/2019 a las 14:21, Teri McElroy escribió:
Hello Lev

I have just recently emailed the list to find out about this same
topic. You can go to the nvda menu, then preferences,
punctuation/symbol pronunciation. You'll need to input the symbols
along with how you want the speech engine to pronounce them. However,
I don't think putting in the unicode values is going to work. You'll
need to input the actual symbol along with how you want NVDA to
pronounce it. I've just tried entering a unicode value here and it
didn't work for me. But when I enter the actual symbol, it does.

You can find information about punctuation/symbol pronunciation in
section 11.2.2 of the online userguide. It will explain about
inputting symbols.

Hope that helps.


Teri


On 13-Mar-19 8:18 AM, Lev wrote:


Hello,

I have noticed that espeak reads some symbols of the international
phonetic alphabet; however, this can not be done by other synthetic
voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of the
synthetizer?

Or, how to make that other synthetizers read those symbols?For
instance, vocalizer.

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

So, is there a file inside the installation folder of nvda or user
settings where I can paste that syntaxis to have nvda reading all
those symbols?

Thanks for any info.

--
Literophilus



enes sarıbaş
 

Hi,

There is an open issue for this. Could you examine the following issue, and perhaps comment?

https://github.com/nvaccess/nvda/issues/2922

On 3/12/2019 7:18 PM, Lev wrote:


 

Hello,

I have noticed that  espeak reads some symbols of the international phonetic alphabet; however, this can not be done by other synthetic voices, so my question is this:

How to make that nvda reads correctly those symbols regardless of the synthetizer?

Or, how to make that other synthetizers read those symbols?  For instance, vocalizer.

 

I got the Unicode transcription for all the symbols that I want.

The syntaxis is this:

U+00E6=Ash

Where the term “ash” is the proper name for that particular symbol.

 

So, is there a file inside the installation folder of nvda or user settings  where I can paste  that syntaxis to have nvda reading all those symbols?

 

Thanks for any info.

 

--
Literophilus


 

Enes,

        The way I read the open issue is that the function of reading a phonetic transcription as it sounds is what's wanted.  It's not 100% clear, but strongly suggested.

        The way I read the request here is to read a phonetic transcription not as it sounds, but broken into atomic units, and not by sound, but by character name of each phoneme (not the sound of the phoneme itself).  Very much more akin to reading words letter by letter, but where each letter is the IPA character name (or description, as vowels are generally described rather than named).
--

Brian - Windows 10 Home, 64-Bit, Version 1809, Build 17763  

A great deal of intelligence can be invested in ignorance when the need for illusion is deep.

          ~ Saul Bellow, To Jerusalem and Back

 

 


Teri McElroy <shimzin.lists@...>
 

Hi Brian

I just wanted to comment on this one. Reading character by character is more fair than having a speech synthesizer be able to read a whole word phonetically. If a blind student is studying linguistics, having their synthesizer be able to read phonetic words would give that student an unfair advantage over sighted students who have to read a phonetic transcription character by character.


Teri

On 15-Mar-19 4:20 PM, Brian Vogel wrote:
Enes,

        The way I read the open issue is that the function of reading a phonetic transcription as it sounds is what's wanted. It's not 100% clear, but strongly suggested.

        The way I read the request here is to read a phonetic transcription not as it sounds, but broken into atomic units, and not by sound, but by character name of each phoneme (not the sound of the phoneme itself).  Very much more akin to reading words letter by letter, but where each letter is the IPA character name (or description, as vowels are generally described rather than named).
--

Brian *-*Windows 10 Home, 64-Bit, Version 1809, Build 17763

*/A great deal of intelligence can be invested in ignorance when the need for illusion is deep./*

          ~ Saul Bellow, /To Jerusalem and Back/


Sarah k Alawami
 

I also read character by character as well when I took my diction for singers class in 2013. Everyone did it and my teacher insisted I do that to. It was hard but I am actually quite surprised I passed the class with a decent grade. But yes reading character by character is almost a must in this class.

On 16 Mar 2019, at 16:48, Teri McElroy wrote:

Hi Brian

I just wanted to comment on this one. Reading character by character is more fair than having a speech synthesizer be able to read a whole word phonetically. If a blind student is studying linguistics, having their synthesizer be able to read phonetic words would give that student an unfair advantage over sighted students who have to read a phonetic transcription character by character.


Teri

On 15-Mar-19 4:20 PM, Brian Vogel wrote:

Enes,

        The way I read the open issue is that the function of reading a phonetic transcription as it sounds is what's wanted. It's not 100% clear, but strongly suggested.

        The way I read the request here is to read a phonetic transcription not as it sounds, but broken into atomic units, and not by sound, but by character name of each phoneme (not the sound of the phoneme itself).  Very much more akin to reading words letter by letter, but where each letter is the IPA character name (or description, as vowels are generally described rather than named).
--

Brian *-*Windows 10 Home, 64-Bit, Version 1809, Build 17763

*/A great deal of intelligence can be invested in ignorance when the need for illusion is deep./*

          ~ Saul Bellow, /To Jerusalem and Back/


 

A couple of things here:

        My comment regarding character by character reading, using character names, of the IPA versus phoneme by phoneme reading of characters in the IPA had to do with the ticket Enes Sarıbaş pointed out on github for NVDA.  It appears to me that ticket is requesting phoneme by phoneme reading, possibly at speeds that would make it impossible to know whether what was being read was originally written in its source language or the IPA.   I was contrasting this with what I interpret Teri's desire to be here, reading a word phonetically transcribed in the IPA character by character, using the name or description for that character, not the actual sound it makes.

        As I posted earlier on a different topic that I spun off just to determine how and where speech dictionaries versus character and symbol stuff is kept:  The symbols & punctuation dictionary is a thing of its own and appears to be localized by language, with the main folder being C:\Program Files (x86)\NVDA\locale\ with all the language abbreviations as separate folders beneath.  The English version being, C:\Program Files (x86)\NVDA\locale\en\symbols.dic.   Here is the section just for the common symbols, as a point of reference with regard to the layout.  The individual fields are tab separated:

# Standard punctuation/symbols
! bang all
" quote most
\# number some
$ dollar all norep
£ pound all norep
euro all norep
¢ Cents all norep
.
.
.
~ tilda mo

The first column is the symbol itself, the second is what you want NVDA to say when it encounters it (its replacement), the third is something called level [which I don't quite understand based on the options, I'm guessing "character" is what will be wanted for the IPA characters, maybe "all"], and the last column is whether to pass the actual character itself to the synthesizer (and I'd presume that would be never for the IPA characters).  Given that every common and not so common symbol I can think of is stored in this file, including all the bullet characters, monetary designations, vulgur fractions, etc., it would seem the IPA would get tacked on in this file.

What's weird, at least to me, is that if you open the character/symbol pronunciation dialog a great deal of what's in the symbols.dic file never appears at all in the list for the Punctuation/Symbol pronunciation dialog where you'd manually enter this stuff.  Only the complex punctuation shows, then it skips right along to showing what I'll call the "emoji characters" which are not stored in the symbols.dic file, but in a file called cldr.dic that's stored in the same folder.   There's a third file in that folder, characterDescriptions.dic, that contains two columns, the first being the letter of the alphabet and the second being its callout word, e.g., A Alpha, B Bravo, C Charlie, etc.

--

Brian - Windows 10 Home, 64-Bit, Version 1809, Build 17763  

A great deal of intelligence can be invested in ignorance when the need for illusion is deep.

          ~ Saul Bellow, To Jerusalem and Back

 

 


 

Correction on what I observed about the Punctuation/Symbol pronunciation dialog.   I'm not sure if I had a system burp or what, but when I got out of NVDA and back in again, and opened it, all the characters from symbols.dic and cldr.dic were present.  It's still odd to me that it puts the complex characters from symbols.dic first, then interrupts the list with the contents from cldr.dic, then comes back to the rest of the content of symbols.dic at the end.

I had to hit the END key to throw myself to the very end of a 2000 plus character list and find more of the contents from symbol.dic when in the character list part of that dialog.
--

Brian - Windows 10 Home, 64-Bit, Version 1809, Build 17763  

A great deal of intelligence can be invested in ignorance when the need for illusion is deep.

          ~ Saul Bellow, To Jerusalem and Back