Issues with some Espeak languages


Kara Goldfinch <kara.louise18@...>
 

Hi all.
This doesn't effect me directly as I only speak English, but I have noticed that there are some major issues with Espeak's support for some languages that don't use Latin characters, particularly Arabic, Burmese and Japanese.
The Arabic and Burmese voices seem to read as though the rules files are missing, just spitting out phonemes and hoping for the best. The Japanese voice just says "Chinese letter" or "Japanese letter" with a sort of Scottish accent when encountering characters.
As far as I know, all other languages that don't use Latin characters aren't affected.
Is this a known issue? If not, maybe perhaps someone who speaks these languages could file a bug report.
I am using NVDA Next build 14212,1c3743ee.
All the best,

Kara Goldfinch

Certified Expert               
View the list of all NVDA Experts


 

hello.
i speak persian.
but i tested arabic and japanese in espeak ng and confirm your message.
also, when i tested japanese language, nvda only told japanese and
chinese letter.
and with japanese language, nvda reads even english message and
document letter by letter, not complete sentence!
also, its arabic is completely ununderstandable.
i realy need arabic free tts and nvda espeak sometimes reads arabic
letter by letter and sometimes very wrong pronunciation and also
sometimes completely ununderstandable words with very bad quality and
not clear!
thanks very very much for your reporting and God bless you!

On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com> wrote:
Hi all.
This doesn't effect me directly as I only speak English, but I have
noticed that there are some major issues with Espeak's support for some
languages that don't use Latin characters, particularly Arabic, Burmese
and Japanese.
The Arabic and Burmese voices seem to read as though the rules files are
missing, just spitting out phonemes and hoping for the best. The
Japanese voice just says "Chinese letter" or "Japanese letter" with a
sort of Scottish accent when encountering characters.
As far as I know, all other languages that don't use Latin characters
aren't affected.
Is this a known issue? If not, maybe perhaps someone who speaks these
languages could file a bug report.
I am using NVDA Next build 14212,1c3743ee.
All the best,

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>
--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org


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

Perhaps you do not have the language pack for that language installed on your machine?
Brian

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

----- Original Message -----
From: "Kara Goldfinch" <kara.louise18@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Thursday, July 20, 2017 10:34 PM
Subject: [nvda] Issues with some Espeak languages


Hi all.
This doesn't effect me directly as I only speak English, but I have
noticed that there are some major issues with Espeak's support for some
languages that don't use Latin characters, particularly Arabic, Burmese
and Japanese.
The Arabic and Burmese voices seem to read as though the rules files are
missing, just spitting out phonemes and hoping for the best. The
Japanese voice just says "Chinese letter" or "Japanese letter" with a
sort of Scottish accent when encountering characters.
As far as I know, all other languages that don't use Latin characters
aren't affected.
Is this a known issue? If not, maybe perhaps someone who speaks these
languages could file a bug report.
I am using NVDA Next build 14212,1c3743ee.
All the best,

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>


Ben J. Bloomgren
 

OK, I didn't realize it was that bad, but yes I have heard of stuff like that. It's atrocious in Vietnamese by the way. In that language, it says the words, but the tones and some of the phonemes themselves are absolutely atrocious. I hardly speak Vietnamese, and I can tell it's absolutely stunningly bad.

On Jul 21, 2017, at 00:53, Brian's Mail list account via Groups.Io <bglists=blueyonder.co.uk@groups.io> wrote:

Perhaps you do not have the language pack for that language installed on your machine?
Brian

bglists@blueyonder.co.uk
Sent via blueyonder.
Please address personal email to:-
briang1@blueyonder.co.uk, putting 'Brian Gaff'
in the display name field.
----- Original Message ----- From: "Kara Goldfinch" <kara.louise18@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Thursday, July 20, 2017 10:34 PM
Subject: [nvda] Issues with some Espeak languages


Hi all.
This doesn't effect me directly as I only speak English, but I have
noticed that there are some major issues with Espeak's support for some
languages that don't use Latin characters, particularly Arabic, Burmese
and Japanese.
The Arabic and Burmese voices seem to read as though the rules files are
missing, just spitting out phonemes and hoping for the best. The
Japanese voice just says "Chinese letter" or "Japanese letter" with a
sort of Scottish accent when encountering characters.
As far as I know, all other languages that don't use Latin characters
aren't affected.
Is this a known issue? If not, maybe perhaps someone who speaks these
languages could file a bug report.
I am using NVDA Next build 14212,1c3743ee.
All the best,

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>



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

There are two distinct issues. Do you have the language pack in the machine, which is what causes the Chinese letter messages etc, and a synth with the right phonemes for other more difficult languages. Often the sounds that Latin languages need are just not enough to create oriental and Arabic ones according to my fiends who speak them.
I suppose this is why you often need a special synth designed to do it correctly.
Brian

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

----- Original Message -----
From: "nasrin khaksar" <nasrinkhaksar3@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Friday, July 21, 2017 6:17 AM
Subject: Re: [nvda] Issues with some Espeak languages


hello.
i speak persian.
but i tested arabic and japanese in espeak ng and confirm your message.
also, when i tested japanese language, nvda only told japanese and
chinese letter.
and with japanese language, nvda reads even english message and
document letter by letter, not complete sentence!
also, its arabic is completely ununderstandable.
i realy need arabic free tts and nvda espeak sometimes reads arabic
letter by letter and sometimes very wrong pronunciation and also
sometimes completely ununderstandable words with very bad quality and
not clear!
thanks very very much for your reporting and God bless you!

On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com> wrote:
Hi all.
This doesn't effect me directly as I only speak English, but I have
noticed that there are some major issues with Espeak's support for some
languages that don't use Latin characters, particularly Arabic, Burmese
and Japanese.
The Arabic and Burmese voices seem to read as though the rules files are
missing, just spitting out phonemes and hoping for the best. The
Japanese voice just says "Chinese letter" or "Japanese letter" with a
sort of Scottish accent when encountering characters.
As far as I know, all other languages that don't use Latin characters
aren't affected.
Is this a known issue? If not, maybe perhaps someone who speaks these
languages could file a bug report.
I am using NVDA Next build 14212,1c3743ee.
All the best,

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>

--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org


 

espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io> wrote:
There are two distinct issues. Do you have the language pack in the machine,

which is what causes the Chinese letter messages etc, and a synth with the
right phonemes for other more difficult languages. Often the sounds that
Latin languages need are just not enough to create oriental and Arabic ones

according to my fiends who speak them.
I suppose this is why you often need a special synth designed to do it
correctly.
Brian

bglists@blueyonder.co.uk
Sent via blueyonder.
Please address personal email to:-
briang1@blueyonder.co.uk, putting 'Brian Gaff'
in the display name field.
----- Original Message -----
From: "nasrin khaksar" <nasrinkhaksar3@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Friday, July 21, 2017 6:17 AM
Subject: Re: [nvda] Issues with some Espeak languages


hello.
i speak persian.
but i tested arabic and japanese in espeak ng and confirm your message.
also, when i tested japanese language, nvda only told japanese and
chinese letter.
and with japanese language, nvda reads even english message and
document letter by letter, not complete sentence!
also, its arabic is completely ununderstandable.
i realy need arabic free tts and nvda espeak sometimes reads arabic
letter by letter and sometimes very wrong pronunciation and also
sometimes completely ununderstandable words with very bad quality and
not clear!
thanks very very much for your reporting and God bless you!

On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com> wrote:
Hi all.
This doesn't effect me directly as I only speak English, but I have
noticed that there are some major issues with Espeak's support for some
languages that don't use Latin characters, particularly Arabic, Burmese
and Japanese.
The Arabic and Burmese voices seem to read as though the rules files are
missing, just spitting out phonemes and hoping for the best. The
Japanese voice just says "Chinese letter" or "Japanese letter" with a
sort of Scottish accent when encountering characters.
As far as I know, all other languages that don't use Latin characters
aren't affected.
Is this a known issue? If not, maybe perhaps someone who speaks these
languages could file a bug report.
I am using NVDA Next build 14212,1c3743ee.
All the best,

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>

--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org






--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org


Quentin Christensen
 

We recently moved to an updated build of eSpeak NG in snapshot builds to fix another issue I can't recall off the top of my head.  If those who've seen this issue could please confirm that it works as expected in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar <nasrinkhaksar3@...> wrote:
espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io> wrote:
> There are two distinct issues. Do you have the language pack in the machine,
>
> which is what causes the Chinese letter messages etc, and a synth with the
> right phonemes for other more difficult languages. Often the sounds that
> Latin languages need are just not enough to create oriental and Arabic ones
>
> according to my fiends who speak them.
>  I suppose this is why you often need a special synth designed to do it
> correctly.
>  Brian
>
> bglists@...
> Sent via blueyonder.
> Please address personal email to:-
> briang1@..., putting 'Brian Gaff'
> in the display name field.
> ----- Original Message -----
> From: "nasrin khaksar" <nasrinkhaksar3@...>
> To: <nvda@nvda.groups.io>
> Sent: Friday, July 21, 2017 6:17 AM
> Subject: Re: [nvda] Issues with some Espeak languages
>
>
>> hello.
>> i speak persian.
>> but i tested arabic and japanese in espeak ng and confirm your message.
>> also, when i tested japanese language, nvda only told japanese and
>> chinese letter.
>> and with japanese language, nvda reads even english message and
>> document letter by letter, not complete sentence!
>> also, its arabic is completely ununderstandable.
>> i realy need arabic free tts and nvda espeak sometimes reads arabic
>> letter by letter and sometimes very wrong pronunciation and also
>> sometimes completely ununderstandable words with very bad quality and
>> not clear!
>> thanks very very much for your reporting and God bless you!
>>
>> On 7/21/17, Kara Goldfinch <kara.louise18@...> wrote:
>>> Hi all.
>>> This doesn't effect me directly as I only speak English, but I have
>>> noticed that there are some major issues with Espeak's support for some
>>> languages that don't use Latin characters, particularly Arabic, Burmese
>>> and Japanese.
>>> The Arabic and Burmese voices seem to read as though the rules files are
>>> missing, just spitting out phonemes and hoping for the best. The
>>> Japanese voice just says "Chinese letter" or "Japanese letter" with a
>>> sort of Scottish accent when encountering characters.
>>> As far as I know, all other languages that don't use Latin characters
>>> aren't affected.
>>> Is this a known issue? If not, maybe perhaps someone who speaks these
>>> languages could file a bug report.
>>> I am using NVDA Next build 14212,1c3743ee.
>>> All the best,
>>>
>>> Kara Goldfinch
>>>
>>> Certified Expert NVDA Certified Expert logo
>>> View the list of all NVDA Experts <https://certification.nvaccess.org/>
>>>
>>
>>
>> --
>> we have not sent you but as a mercy to the creation.
>> holy quran, chapter 21, verse 107.
>> in the very authentic narration is:
>> imam hosein is the beacon of light and the ark of salvation.
>> best website for studying islamic book in different languages
>> al-islam.org
>>
>>
>>
>
>
>
>
>


--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org






--
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now available: http://www.nvaccess.org/shop/

www.nvaccess.org 
Facebook: http://www.facebook.com/NVAccess 
Twitter: @NVAccess 


Kara Goldfinch <kara.louise18@...>
 

Hi Quentin,

Similar issues appear in 2017.2. The only changes being that the Japanese voice says Japanese or Chinese letter using the British English voice, and the Burmese one reads what I Assume are the Unicode character values rather than stringing random phonemes together.

Brian, I have Firefox set to display Unicode characters so they do display correctly. If not, they would display as random symbols. 


Kara

On 21/07/2017 13:52, Quentin Christensen wrote:
We recently moved to an updated build of eSpeak NG in snapshot builds to fix another issue I can't recall off the top of my head.  If those who've seen this issue could please confirm that it works as expected in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar <nasrinkhaksar3@...> wrote:
espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io> wrote:
> There are two distinct issues. Do you have the language pack in the machine,
>
> which is what causes the Chinese letter messages etc, and a synth with the
> right phonemes for other more difficult languages. Often the sounds that
> Latin languages need are just not enough to create oriental and Arabic ones
>
> according to my fiends who speak them.
>  I suppose this is why you often need a special synth designed to do it
> correctly.
>  Brian
>
> bglists@...
> Sent via blueyonder.
> Please address personal email to:-
> briang1@..., putting 'Brian Gaff'
> in the display name field.
> ----- Original Message -----
> From: "nasrin khaksar" <nasrinkhaksar3@...>
> To: <nvda@nvda.groups.io>
> Sent: Friday, July 21, 2017 6:17 AM
> Subject: Re: [nvda] Issues with some Espeak languages
>
>
>> hello.
>> i speak persian.
>> but i tested arabic and japanese in espeak ng and confirm your message.
>> also, when i tested japanese language, nvda only told japanese and
>> chinese letter.
>> and with japanese language, nvda reads even english message and
>> document letter by letter, not complete sentence!
>> also, its arabic is completely ununderstandable.
>> i realy need arabic free tts and nvda espeak sometimes reads arabic
>> letter by letter and sometimes very wrong pronunciation and also
>> sometimes completely ununderstandable words with very bad quality and
>> not clear!
>> thanks very very much for your reporting and God bless you!
>>
>> On 7/21/17, Kara Goldfinch <kara.louise18@...> wrote:
>>> Hi all.
>>> This doesn't effect me directly as I only speak English, but I have
>>> noticed that there are some major issues with Espeak's support for some
>>> languages that don't use Latin characters, particularly Arabic, Burmese
>>> and Japanese.
>>> The Arabic and Burmese voices seem to read as though the rules files are
>>> missing, just spitting out phonemes and hoping for the best. The
>>> Japanese voice just says "Chinese letter" or "Japanese letter" with a
>>> sort of Scottish accent when encountering characters.
>>> As far as I know, all other languages that don't use Latin characters
>>> aren't affected.
>>> Is this a known issue? If not, maybe perhaps someone who speaks these
>>> languages could file a bug report.
>>> I am using NVDA Next build 14212,1c3743ee.
>>> All the best,
>>>
>>> Kara Goldfinch
>>>
>>> Certified Expert NVDA Certified Expert logo
>>> View the list of all NVDA Experts <https://certification.nvaccess.org/>
>>>
>>
>>
>> --
>> we have not sent you but as a mercy to the creation.
>> holy quran, chapter 21, verse 107.
>> in the very authentic narration is:
>> imam hosein is the beacon of light and the ark of salvation.
>> best website for studying islamic book in different languages
>> al-islam.org
>>
>>
>>
>
>
>
>
>


--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org






--
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now available: http://www.nvaccess.org/shop/

www.nvaccess.org 
Facebook: http://www.facebook.com/NVAccess 
Twitter: @NVAccess 


--

Kara Goldfinch

Certified Expert               
View the list of all NVDA Experts


 

hi quentin.
i tested nvda 2017.2 and the issues which i mentioned in my first
comment is unfortunately persists!
very aweful quality, cant understand arabic sentences and espeak says
most arabic words letter by letter.
how can i fix it and study my arabic books?
God bless you and i pray sincerely every day for you.

On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com> wrote:
Hi Quentin,

Similar issues appear in 2017.2. The only changes being that the
Japanese voice says Japanese or Chinese letter using the British English
voice, and the Burmese one reads what I Assume are the Unicode character
values rather than stringing random phonemes together.

Brian, I have Firefox set to display Unicode characters so they do
display correctly. If not, they would display as random symbols.


Kara
On 21/07/2017 13:52, Quentin Christensen wrote:
We recently moved to an updated build of eSpeak NG in snapshot builds
to fix another issue I can't recall off the top of my head. If those
who've seen this issue could please confirm that it works as expected
in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar
<nasrinkhaksar3@gmail.com <mailto:nasrinkhaksar3@gmail.com>> wrote:

espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io
<mailto:blueyonder.co.uk@groups.io>> wrote:
> There are two distinct issues. Do you have the language pack in
the machine,
>
> which is what causes the Chinese letter messages etc, and a
synth with the
> right phonemes for other more difficult languages. Often the
sounds that
> Latin languages need are just not enough to create oriental and
Arabic ones
>
> according to my fiends who speak them.
> I suppose this is why you often need a special synth designed
to do it
> correctly.
> Brian
>
> bglists@blueyonder.co.uk <mailto:bglists@blueyonder.co.uk>
> Sent via blueyonder.
> Please address personal email to:-
> briang1@blueyonder.co.uk <mailto:briang1@blueyonder.co.uk>,
putting 'Brian Gaff'
> in the display name field.
> ----- Original Message -----
> From: "nasrin khaksar" <nasrinkhaksar3@gmail.com
<mailto:nasrinkhaksar3@gmail.com>>
> To: <nvda@nvda.groups.io <mailto:nvda@nvda.groups.io>>
> Sent: Friday, July 21, 2017 6:17 AM
> Subject: Re: [nvda] Issues with some Espeak languages
>
>
>> hello.
>> i speak persian.
>> but i tested arabic and japanese in espeak ng and confirm your
message.
>> also, when i tested japanese language, nvda only told japanese and
>> chinese letter.
>> and with japanese language, nvda reads even english message and
>> document letter by letter, not complete sentence!
>> also, its arabic is completely ununderstandable.
>> i realy need arabic free tts and nvda espeak sometimes reads
arabic
>> letter by letter and sometimes very wrong pronunciation and also
>> sometimes completely ununderstandable words with very bad
quality and
>> not clear!
>> thanks very very much for your reporting and God bless you!
>>
>> On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com
<mailto:kara.louise18@gmail.com>> wrote:
>>> Hi all.
>>> This doesn't effect me directly as I only speak English, but I
have
>>> noticed that there are some major issues with Espeak's support
for some
>>> languages that don't use Latin characters, particularly
Arabic, Burmese
>>> and Japanese.
>>> The Arabic and Burmese voices seem to read as though the rules
files are
>>> missing, just spitting out phonemes and hoping for the best. The
>>> Japanese voice just says "Chinese letter" or "Japanese letter"
with a
>>> sort of Scottish accent when encountering characters.
>>> As far as I know, all other languages that don't use Latin
characters
>>> aren't affected.
>>> Is this a known issue? If not, maybe perhaps someone who
speaks these
>>> languages could file a bug report.
>>> I am using NVDA Next build 14212,1c3743ee.
>>> All the best,
>>>
>>> Kara Goldfinch
>>>
>>> Certified Expert NVDA Certified Expert logo
>>> View the list of all NVDA Experts
<https://certification.nvaccess.org/
<https://certification.nvaccess.org/>>
>>>
>>
>>
>> --
>> we have not sent you but as a mercy to the creation.
>> holy quran, chapter 21, verse 107.
>> in the very authentic narration is:
>> imam hosein is the beacon of light and the ark of salvation.
>> best website for studying islamic book in different languages
>> al-islam.org <http://al-islam.org>
>>
>>
>>
>
>
>
>
>


--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org <http://al-islam.org>






--
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now
available: http://www.nvaccess.org/shop/

Ph +61 7 3149 3306 <tel:%2B61%207%203149%203306>
www.nvaccess.org <http://www.nvaccess.org/>
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess
--

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>
--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org


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

I've just been having some issues with the English Espeak today. In some pdfs and word documents some words get spoken as if the language of the synth has suddenly changed to French or Spanish, so normal words read on other synths sound well foreign. If only I could come up with a sure fire way of demonstrating this, I'd raise a ticket, as it may well be related in that if some code in the file is changing the actual way nvda is interacting with characters when its not supposed to, its going to create problems. of course it could be that nvda is outputting illegal characters to the synths and some just ignore these but espeak uses them. This might be something as simple as the bit mask for changes in a variable is not correct so higher bits are read in inappropriate situations.
Brian

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

----- Original Message -----
From: "Kara Goldfinch" <kara.louise18@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Friday, July 21, 2017 4:01 PM
Subject: Re: [nvda] Issues with some Espeak languages


Hi Quentin,

Similar issues appear in 2017.2. The only changes being that the
Japanese voice says Japanese or Chinese letter using the British English
voice, and the Burmese one reads what I Assume are the Unicode character
values rather than stringing random phonemes together.

Brian, I have Firefox set to display Unicode characters so they do
display correctly. If not, they would display as random symbols.


Kara
On 21/07/2017 13:52, Quentin Christensen wrote:
We recently moved to an updated build of eSpeak NG in snapshot builds
to fix another issue I can't recall off the top of my head. If those
who've seen this issue could please confirm that it works as expected
in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar
<nasrinkhaksar3@gmail.com <mailto:nasrinkhaksar3@gmail.com>> wrote:

espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io
<mailto:blueyonder.co.uk@groups.io>> wrote:
> There are two distinct issues. Do you have the language pack in
the machine,
>
> which is what causes the Chinese letter messages etc, and a
synth with the
> right phonemes for other more difficult languages. Often the
sounds that
> Latin languages need are just not enough to create oriental and
Arabic ones
>
> according to my fiends who speak them.
> I suppose this is why you often need a special synth designed
to do it
> correctly.
> Brian
>
> bglists@blueyonder.co.uk <mailto:bglists@blueyonder.co.uk>
> Sent via blueyonder.
> Please address personal email to:-
> briang1@blueyonder.co.uk <mailto:briang1@blueyonder.co.uk>,
putting 'Brian Gaff'
> in the display name field.
> ----- Original Message -----
> From: "nasrin khaksar" <nasrinkhaksar3@gmail.com
<mailto:nasrinkhaksar3@gmail.com>>
> To: <nvda@nvda.groups.io <mailto:nvda@nvda.groups.io>>
> Sent: Friday, July 21, 2017 6:17 AM
> Subject: Re: [nvda] Issues with some Espeak languages
>
>
>> hello.
>> i speak persian.
>> but i tested arabic and japanese in espeak ng and confirm your
message.
>> also, when i tested japanese language, nvda only told japanese and
>> chinese letter.
>> and with japanese language, nvda reads even english message and
>> document letter by letter, not complete sentence!
>> also, its arabic is completely ununderstandable.
>> i realy need arabic free tts and nvda espeak sometimes reads
arabic
>> letter by letter and sometimes very wrong pronunciation and also
>> sometimes completely ununderstandable words with very bad
quality and
>> not clear!
>> thanks very very much for your reporting and God bless you!
>>
>> On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com
<mailto:kara.louise18@gmail.com>> wrote:
>>> Hi all.
>>> This doesn't effect me directly as I only speak English, but I
have
>>> noticed that there are some major issues with Espeak's support
for some
>>> languages that don't use Latin characters, particularly
Arabic, Burmese
>>> and Japanese.
>>> The Arabic and Burmese voices seem to read as though the rules
files are
>>> missing, just spitting out phonemes and hoping for the best. The
>>> Japanese voice just says "Chinese letter" or "Japanese letter"
with a
>>> sort of Scottish accent when encountering characters.
>>> As far as I know, all other languages that don't use Latin
characters
>>> aren't affected.
>>> Is this a known issue? If not, maybe perhaps someone who
speaks these
>>> languages could file a bug report.
>>> I am using NVDA Next build 14212,1c3743ee.
>>> All the best,
>>>
>>> Kara Goldfinch
>>>
>>> Certified Expert NVDA Certified Expert logo
>>> View the list of all NVDA Experts
<https://certification.nvaccess.org/
<https://certification.nvaccess.org/>>
>>>
>>
>>
>> --
>> we have not sent you but as a mercy to the creation.
>> holy quran, chapter 21, verse 107.
>> in the very authentic narration is:
>> imam hosein is the beacon of light and the ark of salvation.
>> best website for studying islamic book in different languages
>> al-islam.org <http://al-islam.org>
>>
>>
>>
>
>
>
>
>


--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org <http://al-islam.org>






--
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now
available: http://www.nvaccess.org/shop/

Ph +61 7 3149 3306 <tel:%2B61%207%203149%203306>
www.nvaccess.org <http://www.nvaccess.org/>
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess
--

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>


 

brian,
do you mean that you have problem using english with espeak for espeak
ng 1.49.1 final release?
because i dont have problem using english with nvda 2017.1 and 2017.2
which support espeak ng developers.

On 7/22/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io> wrote:
I've just been having some issues with the English Espeak today. In some
pdfs and word documents some words get spoken as if the language of the
synth has suddenly changed to French or Spanish, so normal words read on
other synths sound well foreign. If only I could come up with a sure fire
way of demonstrating this, I'd raise a ticket, as it may well be related in

that if some code in the file is changing the actual way nvda is interacting

with characters when its not supposed to, its going to create problems. of
course it could be that nvda is outputting illegal characters to the synths

and some just ignore these but espeak uses them. This might be something as

simple as the bit mask for changes in a variable is not correct so higher
bits are read in inappropriate situations.
Brian

bglists@blueyonder.co.uk
Sent via blueyonder.
Please address personal email to:-
briang1@blueyonder.co.uk, putting 'Brian Gaff'
in the display name field.
----- Original Message -----
From: "Kara Goldfinch" <kara.louise18@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Friday, July 21, 2017 4:01 PM
Subject: Re: [nvda] Issues with some Espeak languages


Hi Quentin,

Similar issues appear in 2017.2. The only changes being that the
Japanese voice says Japanese or Chinese letter using the British English
voice, and the Burmese one reads what I Assume are the Unicode character
values rather than stringing random phonemes together.

Brian, I have Firefox set to display Unicode characters so they do
display correctly. If not, they would display as random symbols.


Kara
On 21/07/2017 13:52, Quentin Christensen wrote:
We recently moved to an updated build of eSpeak NG in snapshot builds
to fix another issue I can't recall off the top of my head. If those
who've seen this issue could please confirm that it works as expected
in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar
<nasrinkhaksar3@gmail.com <mailto:nasrinkhaksar3@gmail.com>> wrote:

espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io
<mailto:blueyonder.co.uk@groups.io>> wrote:
> There are two distinct issues. Do you have the language pack in
the machine,
>
> which is what causes the Chinese letter messages etc, and a
synth with the
> right phonemes for other more difficult languages. Often the
sounds that
> Latin languages need are just not enough to create oriental and
Arabic ones
>
> according to my fiends who speak them.
> I suppose this is why you often need a special synth designed
to do it
> correctly.
> Brian
>
> bglists@blueyonder.co.uk <mailto:bglists@blueyonder.co.uk>
> Sent via blueyonder.
> Please address personal email to:-
> briang1@blueyonder.co.uk <mailto:briang1@blueyonder.co.uk>,
putting 'Brian Gaff'
> in the display name field.
> ----- Original Message -----
> From: "nasrin khaksar" <nasrinkhaksar3@gmail.com
<mailto:nasrinkhaksar3@gmail.com>>
> To: <nvda@nvda.groups.io <mailto:nvda@nvda.groups.io>>
> Sent: Friday, July 21, 2017 6:17 AM
> Subject: Re: [nvda] Issues with some Espeak languages
>
>
>> hello.
>> i speak persian.
>> but i tested arabic and japanese in espeak ng and confirm your
message.
>> also, when i tested japanese language, nvda only told japanese
and
>> chinese letter.
>> and with japanese language, nvda reads even english message and
>> document letter by letter, not complete sentence!
>> also, its arabic is completely ununderstandable.
>> i realy need arabic free tts and nvda espeak sometimes reads
arabic
>> letter by letter and sometimes very wrong pronunciation and also
>> sometimes completely ununderstandable words with very bad
quality and
>> not clear!
>> thanks very very much for your reporting and God bless you!
>>
>> On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com
<mailto:kara.louise18@gmail.com>> wrote:
>>> Hi all.
>>> This doesn't effect me directly as I only speak English, but I
have
>>> noticed that there are some major issues with Espeak's support
for some
>>> languages that don't use Latin characters, particularly
Arabic, Burmese
>>> and Japanese.
>>> The Arabic and Burmese voices seem to read as though the rules
files are
>>> missing, just spitting out phonemes and hoping for the best. The
>>> Japanese voice just says "Chinese letter" or "Japanese letter"
with a
>>> sort of Scottish accent when encountering characters.
>>> As far as I know, all other languages that don't use Latin
characters
>>> aren't affected.
>>> Is this a known issue? If not, maybe perhaps someone who
speaks these
>>> languages could file a bug report.
>>> I am using NVDA Next build 14212,1c3743ee.
>>> All the best,
>>>
>>> Kara Goldfinch
>>>
>>> Certified Expert NVDA Certified Expert logo
>>> View the list of all NVDA Experts
<https://certification.nvaccess.org/
<https://certification.nvaccess.org/>>
>>>
>>
>>
>> --
>> we have not sent you but as a mercy to the creation.
>> holy quran, chapter 21, verse 107.
>> in the very authentic narration is:
>> imam hosein is the beacon of light and the ark of salvation.
>> best website for studying islamic book in different languages
>> al-islam.org <http://al-islam.org>
>>
>>
>>
>
>
>
>
>


--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org <http://al-islam.org>






--
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now
available: http://www.nvaccess.org/shop/

Ph +61 7 3149 3306 <tel:%2B61%207%203149%203306>
www.nvaccess.org <http://www.nvaccess.org/>
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess
--

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>




--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org


Kara Goldfinch <kara.louise18@...>
 

It sounds to me like someone's marked up the document or paragraphs to be in those languages, and NVDA's picking up on it because of the "automatic language switching" option in voice settings.

Try turning that option off and reading them again.


Kara

On 22/07/2017 10:51, Brian's Mail list account via Groups.Io wrote:
I've just been having some issues with the English Espeak today. In some pdfs and word documents some words get spoken as if the language of the synth has suddenly changed to French or Spanish, so normal words read  on other synths sound well foreign. If only I could come up with a sure fire way of demonstrating this, I'd raise a ticket, as it may well be related in that if some code in the file is changing the actual way nvda is interacting with characters when its not supposed to, its going to create problems. of course it could be that nvda is outputting illegal characters to the synths and some just ignore these but espeak uses them. This might be something as simple as the bit mask for changes in a variable is not correct so higher bits are read in inappropriate  situations.
Brian

bglists@...
Sent via blueyonder.
Please address personal email to:-
briang1@..., putting 'Brian Gaff'
in the display name field.
----- Original Message ----- From: "Kara Goldfinch" <kara.louise18@...>
To: <nvda@nvda.groups.io>
Sent: Friday, July 21, 2017 4:01 PM
Subject: Re: [nvda] Issues with some Espeak languages


Hi Quentin,

Similar issues appear in 2017.2. The only changes being that the
Japanese voice says Japanese or Chinese letter using the British English
voice, and the Burmese one reads what I Assume are the Unicode character
values rather than stringing random phonemes together.

Brian, I have Firefox set to display Unicode characters so they do
display correctly. If not, they would display as random symbols.


Kara
On 21/07/2017 13:52, Quentin Christensen wrote:
We recently moved to an updated build of eSpeak NG in snapshot builds
to fix another issue I can't recall off the top of my head.  If those
who've seen this issue could please confirm that it works as expected
in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar
<nasrinkhaksar3@... <mailto:nasrinkhaksar3@...>> wrote:

    espeak is the best synthesizer and i wish that it supports all
    languages without problem!

    On 7/21/17, Brian's Mail list account via Groups.Io
    <bglists@...
    <mailto:blueyonder.co.uk@groups.io>> wrote:
    > There are two distinct issues. Do you have the language pack in
    the machine,
    >
    > which is what causes the Chinese letter messages etc, and a
    synth with the
    > right phonemes for other more difficult languages. Often the
    sounds that
    > Latin languages need are just not enough to create oriental and
    Arabic ones
    >
    > according to my fiends who speak them.
    >  I suppose this is why you often need a special synth designed
    to do it
    > correctly.
    >  Brian
    >
    > bglists@... <mailto:bglists@...>
    > Sent via blueyonder.
    > Please address personal email to:-
    > briang1@... <mailto:briang1@...>,
    putting 'Brian Gaff'
    > in the display name field.
    > ----- Original Message -----
    > From: "nasrin khaksar" <nasrinkhaksar3@...
    <mailto:nasrinkhaksar3@...>>
    > To: <nvda@nvda.groups.io <mailto:nvda@nvda.groups.io>>
    > Sent: Friday, July 21, 2017 6:17 AM
    > Subject: Re: [nvda] Issues with some Espeak languages
    >
    >
    >> hello.
    >> i speak persian.
    >> but i tested arabic and japanese in espeak ng and confirm your
    message.
    >> also, when i tested japanese language, nvda only told japanese and
    >> chinese letter.
    >> and with japanese language, nvda reads even english message and
    >> document letter by letter, not complete sentence!
    >> also, its arabic is completely ununderstandable.
    >> i realy need arabic free tts and nvda espeak sometimes reads arabic
    >> letter by letter and sometimes very wrong pronunciation and also
    >> sometimes completely ununderstandable words with very bad
    quality and
    >> not clear!
    >> thanks very very much for your reporting and God bless you!
    >>
    >> On 7/21/17, Kara Goldfinch <kara.louise18@...
    <mailto:kara.louise18@...>> wrote:
    >>> Hi all.
    >>> This doesn't effect me directly as I only speak English, but I
    have
    >>> noticed that there are some major issues with Espeak's support
    for some
    >>> languages that don't use Latin characters, particularly
    Arabic, Burmese
    >>> and Japanese.
    >>> The Arabic and Burmese voices seem to read as though the rules
    files are
    >>> missing, just spitting out phonemes and hoping for the best. The
    >>> Japanese voice just says "Chinese letter" or "Japanese letter"
    with a
    >>> sort of Scottish accent when encountering characters.
    >>> As far as I know, all other languages that don't use Latin
    characters
    >>> aren't affected.
    >>> Is this a known issue? If not, maybe perhaps someone who
    speaks these
    >>> languages could file a bug report.
    >>> I am using NVDA Next build 14212,1c3743ee.
    >>> All the best,
    >>>
    >>> Kara Goldfinch
    >>>
    >>> Certified Expert NVDA Certified Expert logo
    >>> View the list of all NVDA Experts
    <https://certification.nvaccess.org/
    <https://certification.nvaccess.org/>>
    >>>
    >>
    >>
    >> --
    >> we have not sent you but as a mercy to the creation.
    >> holy quran, chapter 21, verse 107.
    >> in the very authentic narration is:
    >> imam hosein is the beacon of light and the ark of salvation.
    >> best website for studying islamic book in different languages
    >> al-islam.org <http://al-islam.org>
    >>
    >>
    >>
    >
    >
    >
    >
    >


    --
    we have not sent you but as a mercy to the creation.
    holy quran, chapter 21, verse 107.
    in the very authentic narration is:
    imam hosein is the beacon of light and the ark of salvation.
    best website for studying islamic book in different languages
    al-islam.org <http://al-islam.org>






-- 
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now
available: http://www.nvaccess.org/shop/

Ph +61 7 3149 3306 <tel:%2B61%207%203149%203306>
www.nvaccess.org <http://www.nvaccess.org/>
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess


-- 

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>






--

Kara Goldfinch

Certified Expert               
View the list of all NVDA Experts


 

hi.
i tried both "automatic language switching" on and off and for me the
result was the same.

On 7/22/17, Kara Goldfinch <kara.louise18@gmail.com> wrote:
It sounds to me like someone's marked up the document or paragraphs to
be in those languages, and NVDA's picking up on it because of the
"automatic language switching" option in voice settings.

Try turning that option off and reading them again.


Kara

On 22/07/2017 10:51, Brian's Mail list account via Groups.Io wrote:
I've just been having some issues with the English Espeak today. In
some pdfs and word documents some words get spoken as if the language
of the synth has suddenly changed to French or Spanish, so normal
words read on other synths sound well foreign. If only I could come
up with a sure fire way of demonstrating this, I'd raise a ticket, as
it may well be related in that if some code in the file is changing
the actual way nvda is interacting with characters when its not
supposed to, its going to create problems. of course it could be that
nvda is outputting illegal characters to the synths and some just
ignore these but espeak uses them. This might be something as simple
as the bit mask for changes in a variable is not correct so higher
bits are read in inappropriate situations.
Brian

bglists@blueyonder.co.uk
Sent via blueyonder.
Please address personal email to:-
briang1@blueyonder.co.uk, putting 'Brian Gaff'
in the display name field.
----- Original Message ----- From: "Kara Goldfinch"
<kara.louise18@gmail.com>
To: <nvda@nvda.groups.io>
Sent: Friday, July 21, 2017 4:01 PM
Subject: Re: [nvda] Issues with some Espeak languages


Hi Quentin,

Similar issues appear in 2017.2. The only changes being that the
Japanese voice says Japanese or Chinese letter using the British English
voice, and the Burmese one reads what I Assume are the Unicode character
values rather than stringing random phonemes together.

Brian, I have Firefox set to display Unicode characters so they do
display correctly. If not, they would display as random symbols.


Kara
On 21/07/2017 13:52, Quentin Christensen wrote:
We recently moved to an updated build of eSpeak NG in snapshot builds
to fix another issue I can't recall off the top of my head. If those
who've seen this issue could please confirm that it works as expected
in NVDA stable 2017.2?

In any case, we'll look into it.

Regards

Quentin.

On Fri, Jul 21, 2017 at 9:57 PM, nasrin khaksar
<nasrinkhaksar3@gmail.com <mailto:nasrinkhaksar3@gmail.com>> wrote:

espeak is the best synthesizer and i wish that it supports all
languages without problem!

On 7/21/17, Brian's Mail list account via Groups.Io
<bglists=blueyonder.co.uk@groups.io
<mailto:blueyonder.co.uk@groups.io>> wrote:
> There are two distinct issues. Do you have the language pack in
the machine,
>
> which is what causes the Chinese letter messages etc, and a
synth with the
> right phonemes for other more difficult languages. Often the
sounds that
> Latin languages need are just not enough to create oriental and
Arabic ones
>
> according to my fiends who speak them.
> I suppose this is why you often need a special synth designed
to do it
> correctly.
> Brian
>
> bglists@blueyonder.co.uk <mailto:bglists@blueyonder.co.uk>
> Sent via blueyonder.
> Please address personal email to:-
> briang1@blueyonder.co.uk <mailto:briang1@blueyonder.co.uk>,
putting 'Brian Gaff'
> in the display name field.
> ----- Original Message -----
> From: "nasrin khaksar" <nasrinkhaksar3@gmail.com
<mailto:nasrinkhaksar3@gmail.com>>
> To: <nvda@nvda.groups.io <mailto:nvda@nvda.groups.io>>
> Sent: Friday, July 21, 2017 6:17 AM
> Subject: Re: [nvda] Issues with some Espeak languages
>
>
>> hello.
>> i speak persian.
>> but i tested arabic and japanese in espeak ng and confirm your
message.
>> also, when i tested japanese language, nvda only told
japanese and
>> chinese letter.
>> and with japanese language, nvda reads even english message and
>> document letter by letter, not complete sentence!
>> also, its arabic is completely ununderstandable.
>> i realy need arabic free tts and nvda espeak sometimes reads
arabic
>> letter by letter and sometimes very wrong pronunciation and also
>> sometimes completely ununderstandable words with very bad
quality and
>> not clear!
>> thanks very very much for your reporting and God bless you!
>>
>> On 7/21/17, Kara Goldfinch <kara.louise18@gmail.com
<mailto:kara.louise18@gmail.com>> wrote:
>>> Hi all.
>>> This doesn't effect me directly as I only speak English, but I
have
>>> noticed that there are some major issues with Espeak's support
for some
>>> languages that don't use Latin characters, particularly
Arabic, Burmese
>>> and Japanese.
>>> The Arabic and Burmese voices seem to read as though the rules
files are
>>> missing, just spitting out phonemes and hoping for the best.
The
>>> Japanese voice just says "Chinese letter" or "Japanese letter"
with a
>>> sort of Scottish accent when encountering characters.
>>> As far as I know, all other languages that don't use Latin
characters
>>> aren't affected.
>>> Is this a known issue? If not, maybe perhaps someone who
speaks these
>>> languages could file a bug report.
>>> I am using NVDA Next build 14212,1c3743ee.
>>> All the best,
>>>
>>> Kara Goldfinch
>>>
>>> Certified Expert NVDA Certified Expert logo
>>> View the list of all NVDA Experts
<https://certification.nvaccess.org/
<https://certification.nvaccess.org/>>
>>>
>>
>>
>> --
>> we have not sent you but as a mercy to the creation.
>> holy quran, chapter 21, verse 107.
>> in the very authentic narration is:
>> imam hosein is the beacon of light and the ark of salvation.
>> best website for studying islamic book in different languages
>> al-islam.org <http://al-islam.org>
>>
>>
>>
>
>
>
>
>


--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org <http://al-islam.org>






--
Quentin Christensen
Training and Support Manager

Basic Training for NVDA & Microsoft Word with NVDA E-Books now
available: http://www.nvaccess.org/shop/

Ph +61 7 3149 3306 <tel:%2B61%207%203149%203306>
www.nvaccess.org <http://www.nvaccess.org/>
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess
--

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>


--

Kara Goldfinch

Certified Expert NVDA Certified Expert logo
View the list of all NVDA Experts <https://certification.nvaccess.org/>
--
we have not sent you but as a mercy to the creation.
holy quran, chapter 21, verse 107.
in the very authentic narration is:
imam hosein is the beacon of light and the ark of salvation.
best website for studying islamic book in different languages
al-islam.org


 
Edited

Hello!
I sometimes use Espeak-NG in Romanian and I noticed that it has a problem. At some words, it doesn't put the accent on the proper silable, but that's not a major problem. Anyway, I don't know if the developers work to improve this language.