Topics

speech refactor things in the threshold snapshot and introduction

zvonimir stanečić, 9a5dsz
 

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Marco Zehe
 

Hi zvonimir,

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Brian's Mail list account
 

You are lucky to get two. In my windows 7 box it is just Espeak, even the native sappi5 and speech platform are not selectable in the synth dialogue at all. I had assumed this was because the job was only half done so to speak, at least it stops you from selecting them, a little worrying that one can in fact attempt to use others.
Brian

bglists@...
Sent via blueyonder.
Please address personal E-mail to:-
briang1@..., putting 'Brian Gaff'
in the display name field.
Newsgroup monitored: alt.comp.blind-users

----- Original Message -----
From: "Marco Zehe" <@marcozehe>
To: <nvda-devel@groups.io>
Sent: Wednesday, June 05, 2019 8:11 AM
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction


Hi zvonimir,

this is totally expected. Threshold contains the refactored speech engine.
And it appears that backwards compatibility is not yet fully restored.
Neither the Vocalizer nor the SAPI 5 driver have been updated to the new
interfaces yet. The only synths that are expected to work are eSpeak NG and
Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <
zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for
Croatian language.

I have an issue with the latest threshold snaps, with the codefactory
vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but
not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir


 

Hey Marco,

I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.

Regards,
Leonard

Op 5 jun. 2019 om 09:11 heeft Marco Zehe <marco.zehe@...> het volgende geschreven:

Hi zvonimir,

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

zvonimir stanečić, 9a5dsz
 

You don’t understand:

Vocalizer on sapi5 works, but the character spelling is problematic.

I don’t know how vocalizer sends characters to the sapi5 itself.

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of Marco Zehe
Sent: Wednesday, June 5, 2019 9:12 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

Hi zvonimir,

 

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

 

Marco

 

 

Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Michael Curran
 

Jamie had already ported sapi5 in his initial work in 2017.


However, I will look into this. Perhaps the characterMode command is now somehow different to pre-speechRefactor.


Mick



On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:
Hey Marco,

I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.

Regards,
Leonard

Op 5 jun. 2019 om 09:11 heeft Marco Zehe <marco.zehe@...> het volgende geschreven:

Hi zvonimir,

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

zvonimir stanečić, 9a5dsz
 

This caracter mode command somehow breaks vocalizer as an NVDA addon *codefactory

Sapi versio, too

It says for example, number followed by the word space

 

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of Michael Curran
Sent: Wednesday, June 5, 2019 10:58 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

Jamie had already ported sapi5 in his initial work in 2017.

 

However, I will look into this. Perhaps the characterMode command is now somehow different to pre-speechRefactor.

 

Mick

 

 

On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:

Hey Marco,

 

I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.

 

Regards,

Leonard


Op 5 jun. 2019 om 09:11 heeft Marco Zehe <marco.zehe@...> het volgende geschreven:

Hi zvonimir,

 

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

 

Marco

 

 

Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

zvonimir stanečić, 9a5dsz
 

You can just use the tts from the link I posted, and test it like a 7 days trial

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of Michael Curran
Sent: Wednesday, June 5, 2019 10:58 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

Jamie had already ported sapi5 in his initial work in 2017.

 

However, I will look into this. Perhaps the characterMode command is now somehow different to pre-speechRefactor.

 

Mick

 

 

On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:

Hey Marco,

 

I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.

 

Regards,

Leonard


Op 5 jun. 2019 om 09:11 heeft Marco Zehe <marco.zehe@...> het volgende geschreven:

Hi zvonimir,

 

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

 

Marco

 

 

Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

 

Hi,

Several things come to my mind:

  • Compatibility restoration: yes and no, I think. There were things removed n Threshold branch that won’t come back.
  • Character spelling issue: it could be anything, but I’m leaning more towards speech function patching.
  • SAPI5: I’m thinking this may have to do with attempts to load something that isn’t there anymore (BoolSynthSetting, for instance, which will cause various speech synthesizers to fail).

 

Cheers,

Joseph

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of zvonimir stanecic, 9a5dsz
Sent: Wednesday, June 5, 2019 2:28 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

You can just use the tts from the link I posted, and test it like a 7 days trial

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of Michael Curran
Sent: Wednesday, June 5, 2019 10:58 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

Jamie had already ported sapi5 in his initial work in 2017.

 

However, I will look into this. Perhaps the characterMode command is now somehow different to pre-speechRefactor.

 

Mick

 

 

On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:

Hey Marco,

 

I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.

 

Regards,

Leonard


Op 5 jun. 2019 om 09:11 heeft Marco Zehe <marco.zehe@...> het volgende geschreven:

Hi zvonimir,

 

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

 

Marco

 

 

Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

zvonimir stanečić, 9a5dsz
 

Hi,

What’s used instead of bool speech settings?

I saw this in one tts which I modified for internal use.

So I need to update it for the speech refactor

 

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of Joseph Lee
Sent: Wednesday, June 5, 2019 4:07 PM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

Hi,

Several things come to my mind:

  • Compatibility restoration: yes and no, I think. There were things removed n Threshold branch that won’t come back.
  • Character spelling issue: it could be anything, but I’m leaning more towards speech function patching.
  • SAPI5: I’m thinking this may have to do with attempts to load something that isn’t there anymore (BoolSynthSetting, for instance, which will cause various speech synthesizers to fail).

 

Cheers,

Joseph

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of zvonimir stanecic, 9a5dsz
Sent: Wednesday, June 5, 2019 2:28 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

You can just use the tts from the link I posted, and test it like a 7 days trial

 

From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of Michael Curran
Sent: Wednesday, June 5, 2019 10:58 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction

 

Jamie had already ported sapi5 in his initial work in 2017.

 

However, I will look into this. Perhaps the characterMode command is now somehow different to pre-speechRefactor.

 

Mick

 

 

On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:

Hey Marco,

 

I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.

 

Regards,

Leonard


Op 5 jun. 2019 om 09:11 heeft Marco Zehe <marco.zehe@...> het volgende geschreven:

Hi zvonimir,

 

this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

 

Marco

 

 

Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@...>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Brian's Mail list account
 

Well I cannot see it in Windows 7 any more unfortunately or speech platform either.
Brian

bglists@...
Sent via blueyonder.
Please address personal E-mail to:-
briang1@..., putting 'Brian Gaff'
in the display name field.
Newsgroup monitored: alt.comp.blind-users

----- Original Message -----
From: "Michael Curran" <mick@...>
To: <nvda-devel@groups.io>
Sent: Wednesday, June 05, 2019 9:58 AM
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction


Jamie had already ported sapi5 in his initial work in 2017.


However, I will look into this. Perhaps the characterMode command is now
somehow different to pre-speechRefactor.


Mick



On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:
Hey Marco,

I was prett# sure that Mick also ported the sapi5 driver. I could be
wrong, though.

Regards,
Leonard

Op 5 jun. 2019 om 09:11 heeft Marco Zehe <@marcozehe
<mailto:@marcozehe>> het volgende geschreven:

Hi zvonimir,

this is totally expected. Threshold contains the refactored speech
engine. And it appears that backwards compatibility is not yet fully
restored. Neither the Vocalizer nor the SAPI 5 driver have been
updated to the new interfaces yet. The only synths that are expected
to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz
<zvonimirek222@... <mailto:zvonimirek222@...>>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators
for Croatian language.

I have an issue with the latest threshold snaps, with the
codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme
names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from
codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Brian's Mail list account
 

I do not hear any errors I just cannot seem to change it. I'll have a play. I think first to turn off all addons though, just in case we are being bamboozled!
Brian

bglists@...
Sent via blueyonder.
Please address personal E-mail to:-
briang1@..., putting 'Brian Gaff'
in the display name field.
Newsgroup monitored: alt.comp.blind-users

----- Original Message -----
From: "Joseph Lee" <@joslee>
To: <nvda-devel@groups.io>
Sent: Wednesday, June 05, 2019 3:07 PM
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction


Hi,

Several things come to my mind:

* Compatibility restoration: yes and no, I think. There were things removed n Threshold branch that won’t come back.
* Character spelling issue: it could be anything, but I’m leaning more towards speech function patching.
* SAPI5: I’m thinking this may have to do with attempts to load something that isn’t there anymore (BoolSynthSetting, for instance, which will cause various speech synthesizers to fail).



Cheers,

Joseph



From: nvda-devel@groups.io <nvda-devel@groups.io> On Behalf Of zvonimir stanecic, 9a5dsz
Sent: Wednesday, June 5, 2019 2:28 AM
To: nvda-devel@groups.io
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction



You can just use the tts from the link I posted, and test it like a 7 days trial



From: nvda-devel@groups.io <mailto:nvda-devel@groups.io> <nvda-devel@groups.io <mailto:nvda-devel@groups.io> > On Behalf Of Michael Curran
Sent: Wednesday, June 5, 2019 10:58 AM
To: nvda-devel@groups.io <mailto:nvda-devel@groups.io>
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction



Jamie had already ported sapi5 in his initial work in 2017.



However, I will look into this. Perhaps the characterMode command is now somehow different to pre-speechRefactor.



Mick





On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:

Hey Marco,



I was prett# sure that Mick also ported the sapi5 driver. I could be wrong, though.



Regards,

Leonard


Op 5 jun. 2019 om 09:11 heeft Marco Zehe <@marcozehe <mailto:@marcozehe> > het volgende geschreven:

Hi zvonimir,



this is totally expected. Threshold contains the refactored speech engine. And it appears that backwards compatibility is not yet fully restored. Neither the Vocalizer nor the SAPI 5 driver have been updated to the new interfaces yet. The only synths that are expected to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).



Marco





Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz <zvonimirek222@... <mailto:zvonimirek222@...> >:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators for Croatian language.

I have an issue with the latest threshold snaps, with the codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Brian's Mail list account
 

Soor this is a rush message got to go out...
Hi its not in fact that the synths do not work but they are buggy and so is the dialogue, for instance this.

DEBUG - core._terminate (20:52:48.826):
Terminating gui
ERROR - gui.terminate (20:52:48.826):
Destroyed but not deleted instance of settings dialog exists: <gui.settings.NVDASettingsDialog object at 0x05581B20>
ERROR - gui.terminate (20:52:48.831):
Destroyed but not deleted instance of settings dialog exists: <gui.settingsDialogs.NVDASettingsDialog object at 0x056F0B70>
DEBUG - gui.settingsDialogs.SettingsDialog._setInstanceDestroyedState (20:52:48.835):
Setting state to destroyed for instance: <gui.settingsDialogs.NVDASettingsDialog object at 0x05581B20>
Current _instances {<gui.settingsDialogs.NVDASettingsDialog object at 0x05581B20>: 1, <gui.settingsDialogs.NVDASettingsDialog object at 0x056F0B70>: 1}
DEBUG - gui.settingsDialogs.SettingsDialog._setInstanceDestroyedState (20:52:48.838):
Setting state to destroyed for instance: <gui.settingsDialogs.NVDASettingsDialog object at 0x056F0B70>
Current _instances {<gui.settingsDialogs.NVDASettingsDialog object at 0x056F0B70>: 1}
DEBUG - fileUtils.FaultTolerantFile (20:52:48.842):
C:\nvda test folder\userConfig\nvda.inirrasyj.tmp
INFO - config.ConfigManager.save (20:52:48.848):

When you find you cannot close the dialogue, also you get error dings when you finally get the change button to operate, then in speech platform it dings between voices and does not switch and it then sort of does the same in sappi five but in sappi 4 the voices say the same as 5 but do not do anything leaving it on the defualt. Very messy and glithing when you try to get that change synth button to do what you want.
Brian

bglists@...
Sent via blueyonder.
Please address personal E-mail to:-
briang1@..., putting 'Brian Gaff'
in the display name field.
Newsgroup monitored: alt.comp.blind-users

----- Original Message -----
From: "Michael Curran" <mick@...>
To: <nvda-devel@groups.io>
Sent: Wednesday, June 05, 2019 9:58 AM
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction


Jamie had already ported sapi5 in his initial work in 2017.


However, I will look into this. Perhaps the characterMode command is now
somehow different to pre-speechRefactor.


Mick



On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:
Hey Marco,

I was prett# sure that Mick also ported the sapi5 driver. I could be
wrong, though.

Regards,
Leonard

Op 5 jun. 2019 om 09:11 heeft Marco Zehe <@marcozehe
<mailto:@marcozehe>> het volgende geschreven:

Hi zvonimir,

this is totally expected. Threshold contains the refactored speech
engine. And it appears that backwards compatibility is not yet fully
restored. Neither the Vocalizer nor the SAPI 5 driver have been
updated to the new interfaces yet. The only synths that are expected
to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz
<zvonimirek222@... <mailto:zvonimirek222@...>>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators
for Croatian language.

I have an issue with the latest threshold snaps, with the
codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme
names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from
codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir

Brian's Mail list account
 

Hi here are the more detailed results of some tests with add ons disabled.

This is an attempt to show the problems in the Threshold snap with synth switching. I have disabled all add ons which makes the change button now behave itself, However the main errors seem to be in the voice selection areas, in that the actual names of voices in the list are usually not the ones that are for the synth in use, and it seems that in Sappi 5 and Speech platform no voice switching is possible. This voice selection error even affects espeak since it has remembered the voices from the last selected synth, through a closing of the dialogues and reopening. I guess it may come right through a restart but I've not tried it. I will selectively edit a log below.


INFO - core.main (08:00:29.010):
NVDA version threshold-17459,ef3ed13f
INFO - core.main (08:00:29.010):
Using Windows version 6.1.7601 service pack 1 workstation
INFO - core.main (08:00:29.010):
Using Python version 2.7.16 (v2.7.16:413a49145e, Mar 4 2019, 01:30:55) [MSC v.1500 32 bit (Intel)]
INFO - core.main (08:00:29.010):
Using comtypes version 1.1.7
INFO - core.main (08:00:29.010):
Using configobj version 5.1.0 with validate version 1.0.1

snip

INFO - core.main (08:00:29.032):
Add-ons are disabled. Restart NVDA to enable them.

snip

DEBUG - core.main (08:00:29.148):
Initializing speech
INFO - synthDrivers.espeak.SynthDriver.__init__ (08:00:29.160):
Using eSpeak NG version 1.49.3 dev
DEBUG - speechDictHandler.SpeechDict.load (08:00:29.194):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\espeak\espeak-English (Great Britain).dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:00:29.194):
9 loaded records.
DEBUG - synthDriverHandler.SynthDriver.loadSettings (08:00:29.194):
Loaded settings for SynthDriver espeak
INFO - synthDriverHandler.setSynth (08:00:29.194):
Loaded synthDriver espeak
INFO - core.main (08:00:29.194):
Using wx version 4.0.3 msw (phoenix) wxWidgets 3.0.5
DEBUG - core.main (08:00:29.196):

snip

I first selected Sappi4, which in this case used the old Via voice synth, but that seemed to work fine except for the known limited options it has of course, so I'll not muddy the waters with it here, but bear in mind the voice read American was selected before backing out.

These were the sort of errors seeen in the main.

Input: kb(desktop):upArrow
DEBUGWARNING - synthDrivers.sapi4.SynthDriverBufSink.IUnknown_Release (08:01:56.680):
ITTSBufNotifySink::Release called too many times by engine
IO - speech.speak (08:01:56.680):
Speaking [LangChangeCommand ('en_US'), u'47']
DEBUGWARNING - synthDrivers.sapi4.SynthDriver.speak (08:01:56.680):
Unsupported speech command: LangChangeCommand ('en_US')

snip

On the way to opening the speech dialogue again to change I also got this.

IO - inputCore.InputManager.executeGesture (08:02:39.987):
Input: kb(desktop):NVDA+n
DEBUG - gui.MainFrame.evaluateUpdatePendingUpdateMenuItemCommand (08:02:40.000):
Error while removing pending update menu item
Traceback (most recent call last):
File "gui\__init__.pyc", line 233, in evaluateUpdatePendingUpdateMenuItemCommand
wxAssertionError: C++ assertion "node" failed at ..\..\src\msw\menu.cpp(890) in wxMenu::DoRemove(): bug in wxMenu::Remove logic
IO - speech.speak (08:02:40.029):
Speaking [LangChangeCommand ('en_US'), u'NVDA menu']

snip

I'm removing the very convlouted log entries while the menue and dialogues are changed here as it is confusing. I've just selected Sappi 5.

IO - speech.speak (08:03:11.644):
Speaking [LangChangeCommand ('en_US'), u'Select Synthesizer dialog']
DEBUGWARNING - synthDrivers.sapi4.SynthDriver.speak (08:03:11.644):
Unsupported speech command: LangChangeCommand ('en_US')
IO - speech.speak (08:03:11.648):
Speaking [LangChangeCommand ('en_US'), u'Synthesizer: combo box Microsoft Speech API version 4 collapsed Alt+s']
DEBUGWARNING - synthDrivers.sapi4.SynthDriverBufSink.IUnknown_Release (08:03:13.680):
ITTSBufNotifySink::Release called too many times by engine
DEBUGWARNING - synthDrivers.sapi4.SynthDriver.speak (08:03:13.697):
Unsupported speech command: LangChangeCommand ('en_US')
IO - inputCore.InputManager.executeGesture (08:03:14.492):
Input: kb(desktop):downArrow
IO - speech.speak (08:03:14.512):
Speaking [LangChangeCommand ('en_US'), u'Microsoft Speech API version 5']
DEBUGWARNING - synthDrivers.sapi4.SynthDriver.speak (08:03:14.513):
Unsupported speech command: LangChangeCommand ('en_US')
IO - inputCore.InputManager.executeGesture (08:03:18.236):
Input: kb(desktop):tab
IO - speech.speak (08:03:18.256):
Speaking [LangChangeCommand ('en_US'), u'Output device: combo box Microsoft Sound Mapper collapsed Alt+d']
DEBUGWARNING - synthDrivers.sapi4.SynthDriver.speak (08:03:18.256):
Unsupported speech command: LangChangeCommand ('en_US')
IO - inputCore.InputManager.executeGesture (08:03:22.427):
Input: kb(desktop):shift+tab
IO - speech.speak (08:03:22.434):
Speaking [LangChangeCommand ('en_US'), u'Synthesizer: combo box Microsoft Speech API version 5 collapsed Alt+s']
DEBUGWARNING - synthDrivers.sapi4.SynthDriver.speak (08:03:22.437):
Unsupported speech command: LangChangeCommand ('en_US')
IO - inputCore.InputManager.executeGesture (08:03:26.227):
Input: kb(desktop):enter
DEBUGWARNING - synthDrivers.sapi4.SynthDriverBufSink.IUnknown_Release (08:03:26.232):
ITTSBufNotifySink::Release called too many times by engine
DEBUG - speechDictHandler.SpeechDict.load (08:03:26.821):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\sapi5\sapi5-ScanSoft Daniel_Full_22kHz.dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:03:26.825):
1 loaded records.
DEBUG - synthDriverHandler.SynthDriver.loadSettings (08:03:26.825):
Loaded settings for SynthDriver sapi5
INFO - synthDriverHandler.setSynth (08:03:26.825):
Loaded synthDriver sapi5

snip

Current _instances {<gui.settingsDialogs.NVDASettingsDialog object at 0x05DAB260>: 0, <gui.settingsDialogs.SynthesizerSelectionDialog object at 0x05D983F0>: 1}
IO - speech.speak (08:03:26.861):
Speaking [LangChangeCommand ('en_GB'), u'NVDA Settings: Speech (normal configuration) dialog']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:26.862):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - speech.speak (08:03:26.907):
Speaking [LangChangeCommand ('en_GB'), u'Speech property page']
IO - speech.speak (08:03:26.911):
Speaking [LangChangeCommand ('en_GB'), u'Synthesizer grouping']
IO - speech.speak (08:03:26.914):
Speaking [LangChangeCommand ('en_GB'), u'Synthesizer edit read only multi line Alt+s']
IO - speech.speak (08:03:26.915):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Speech API version 5\r']
DEBUG - gui.settingsDialogs.SettingsDialog._setInstanceDestroyedState (08:03:26.934):
Setting state to destroyed for instance: <gui.settingsDialogs.SynthesizerSelectionDialog object at 0x05D983F0>
Current _instances {<gui.settingsDialogs.NVDASettingsDialog object at 0x05DAB260>: 0, <gui.settingsDialogs.SynthesizerSelectionDialog object at 0x05D983F0>: 1}
IO - inputCore.InputManager.executeGesture (08:03:30.788):
Input: kb(desktop):tab
IO - speech.speak (08:03:30.802):
Speaking [LangChangeCommand ('en_GB'), u'Change... button Alt+h']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:30.803):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:03:32.036):
Input: kb(desktop):tab
IO - speech.speak (08:03:32.154):
Speaking [LangChangeCommand ('en_GB'), u'Voice: combo box English-American: Reed (Adult Male) - ViaVoice Outloud 4.0 collapsed Alt+v']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:32.155):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:03:35.963):
Input: kb(desktop):downArrow
DEBUG - speechDictHandler.SpeechDict.load (08:03:36.073):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Reed-Tel (Adult Male for Telephone) - ViaVoice Outloud 4.0.dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:03:36.073):
file '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Reed-Tel (Adult Male for Telephone) - ViaVoice Outloud 4.0.dic' not found.
IO - speech.speak (08:03:36.085):
Speaking [LangChangeCommand ('en_GB'), u'English-American: Reed-Tel (Adult Male for Telephone) - ViaVoice Outloud 4.0']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:36.085):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:03:37.875):
Input: kb(desktop):downArrow
DEBUG - speechDictHandler.SpeechDict.load (08:03:37.983):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Sandy (Child) - ViaVoice Outloud 4.0.dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:03:37.983):
file '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Sandy (Child) - ViaVoice Outloud 4.0.dic' not found.
IO - speech.speak (08:03:37.992):
Speaking [LangChangeCommand ('en_GB'), u'English-American: Sandy (Child) - ViaVoice Outloud 4.0']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:37.993):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:03:40.763):
Input: kb(desktop):downArrow
DEBUG - speechDictHandler.SpeechDict.load (08:03:40.871):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Shelley (Adult Female) - ViaVoice Outloud 4.0.dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:03:40.871):
file '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Shelley (Adult Female) - ViaVoice Outloud 4.0.dic' not found.
IO - speech.speak (08:03:40.880):
Speaking [LangChangeCommand ('en_GB'), u'English-American: Shelley (Adult Female) - ViaVoice Outloud 4.0']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:40.881):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:03:43.194):
Input: kb(desktop):downArrow
DEBUG - speechDictHandler.SpeechDict.load (08:03:43.305):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Shelley-Tel (Adult Female for Telephone) - ViaVoice Outloud 4.0.dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:03:43.305):
file '.\userConfig\speechDicts\voiceDicts.v1\sapi4\sapi4-English-American_ Shelley-Tel (Adult Female for Telephone) - ViaVoice Outloud 4.0.dic' not found.
IO - speech.speak (08:03:43.315):
Speaking [LangChangeCommand ('en_GB'), u'English-American: Shelley-Tel (Adult Female for Telephone) - ViaVoice Outloud 4.0']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:03:43.315):
snip
Well its going for the wrong list as it is still thinking it is on Sappi 4
This I think is where a lot of the non speaking and errors come from. The same is true for other synths as well althoug you can hear the voices change ie for the synth, you cannot in fact select the various voices as the list is wrong.
snip
IO - speech.speak (08:04:54.845):
Speaking [LangChangeCommand ('en_GB'), u'Synthesizer grouping']
IO - speech.speak (08:04:54.846):
Speaking [LangChangeCommand ('en_GB'), u'Synthesizer edit read only multi line Alt+s']
IO - speech.speak (08:04:54.848):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Speech API version 5\r']
IO - inputCore.InputManager.executeGesture (08:04:55.788):
Input: kb(desktop):tab
IO - speech.speak (08:04:55.815):
Speaking [LangChangeCommand ('en_GB'), u'Change... button Alt+h']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:04:55.815):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:04:58.611):
Input: kb(desktop):space
IO - speech.speak (08:04:58.627):
Speaking [LangChangeCommand ('en_GB'), u'space', EndUtteranceCommand()]
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:04:58.627):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - speech.speak (08:04:58.631):
Speaking [LangChangeCommand ('en_GB'), u'pressed']
DEBUG - gui.settingsDialogs.__new__ (08:04:58.716):
Creating new settings dialog (multiInstanceAllowed:True). State of _instances {<gui.settingsDialogs.NVDASettingsDialog object at 0x05D849E0>: 0}
DEBUG - windowUtils.getWindowScalingFactor (08:04:58.717):
GetDpiForWindow failed, using GetDeviceCaps instead
DEBUGWARNING - synthDriverHandler.getSynthList (08:04:58.719):
Synthesizer 'oneCore' doesn't pass the check, excluding from list
IO - speech.speak (08:04:58.878):
Speaking [LangChangeCommand ('en_GB'), u'Select Synthesizer dialog']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:04:58.878):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - speech.speak (08:04:58.884):
Speaking [LangChangeCommand ('en_GB'), u'Synthesizer: combo box Microsoft Speech API version 5 collapsed Alt+s']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:05:00.500):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:05:02.028):
Input: kb(desktop):downArrow
IO - speech.speak (08:05:02.142):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Speech Platform']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:05:02.142):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:05:04.763):
Input: kb(desktop):enter
DEBUG - speechDictHandler.SpeechDict.load (08:05:04.953):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\mssp\mssp-Microsoft Server Speech Text to Speech Voice (en-GB, Hazel).dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:05:04.957):
4 loaded records.
DEBUG - synthDriverHandler.SynthDriver.loadSettings (08:05:04.957):
Loaded settings for SynthDriver mssp
INFO - synthDriverHandler.setSynth (08:05:04.957):
Loaded synthDriver mssp

snip snip I'm going to cut out all the stuff about going to the voice settings and just show you what we get.

Remeber you cannot hear any voice change.

IO - inputCore.InputManager.executeGesture (08:05:09.500):
Input: kb(desktop):tab
IO - speech.speak (08:05:09.523):
Speaking [LangChangeCommand ('en_GB'), u'Voice: combo box ScanSoft Daniel_Full_22kHz collapsed Alt+v']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:05:09.525):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:05:13.171):
Input: kb(desktop):downArrow
ERROR - unhandled exception (08:05:13.176):
Traceback (most recent call last):
File "gui\settingsDialogs.pyc", line 1012, in __call__
File "synthDriverHandler.pyc", line 33, in changeVoice
File "synthDrivers\sapi5.pyc", line 220, in _set_voice
File "synthDrivers\sapi5.pyc", line 165, in _getVoiceTokens
AttributeError: 'NoneType' object has no attribute 'getVoices'
IO - speech.speak (08:05:13.184):
Speaking [LangChangeCommand ('en_GB'), u'ScanSoft Emily_Dri40_16kHz']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:05:13.186):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:05:14.835):
Input: kb(desktop):downArrow
ERROR - unhandled exception (08:05:14.842):
Traceback (most recent call last):
File "gui\settingsDialogs.pyc", line 1012, in __call__
File "synthDriverHandler.pyc", line 33, in changeVoice
File "synthDrivers\sapi5.pyc", line 220, in _set_voice
File "synthDrivers\sapi5.pyc", line 165, in _getVoiceTokens
AttributeError: 'NoneType' object has no attribute 'getVoices'
IO - speech.speak (08:05:14.845):
Speaking [LangChangeCommand ('en_GB'), u'ScanSoft Serena_Full_22kHz']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:05:14.846):
Unsupported speech command: LangChangeCommand ('en_GB')

snip Now when you come out and its using the speech platform, something thinks its still sappi 5 so the voice selection then fails again.
Current _instances {<gui.settingsDialogs.SynthesizerSelectionDialog object at 0x05B6C6C0>: 1, <gui.settingsDialogs.NVDASettingsDialog object at 0x05C749E0>: 0}
IO - inputCore.InputManager.executeGesture (08:07:32.693):
Input: kb(desktop):tab
IO - speech.speak (08:07:32.716):
Speaking [LangChangeCommand ('en_GB'), u'Change... button Alt+h']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:07:32.716):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:07:33.835):
Input: kb(desktop):tab
IO - speech.speak (08:07:33.858):
Speaking [LangChangeCommand ('en_GB'), u'Voice: combo box Microsoft Server Speech Text to Speech Voice (en-GB, Hazel) collapsed Alt+v']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:07:33.859):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:07:37.444):
Input: kb(desktop):downArrow
IO - inputCore.InputManager.executeGesture (08:07:38.180):
Input: kb(desktop):upArrow
ERROR - unhandled exception (08:07:38.180):
Traceback (most recent call last):
File "gui\settingsDialogs.pyc", line 1012, in __call__
File "synthDriverHandler.pyc", line 33, in changeVoice
File "synthDrivers\sapi5.pyc", line 220, in _set_voice
File "synthDrivers\sapi5.pyc", line 165, in _getVoiceTokens
AttributeError: 'NoneType' object has no attribute 'getVoices'
IO - speech.speak (08:07:38.196):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Server Speech Text to Speech Voice (en-US, ZiraPro)']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:07:38.197):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:07:41.332):
Input: kb(desktop):upArrow
IO - inputCore.InputManager.executeGesture (08:07:42.595):
Input: kb(desktop):downArrow
ERROR - unhandled exception (08:07:42.596):
Traceback (most recent call last):
File "gui\settingsDialogs.pyc", line 1012, in __call__
File "synthDriverHandler.pyc", line 33, in changeVoice
File "synthDrivers\sapi5.pyc", line 220, in _set_voice
File "synthDrivers\sapi5.pyc", line 165, in _getVoiceTokens
AttributeError: 'NoneType' object has no attribute 'getVoices'
IO - speech.speak (08:07:42.609):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Server Speech Text to Speech Voice (en-GB, Hazel)']

snip going for the incorrect voice list obviously. So lets skip all the way to Espeak and set that and look at the voices.
IO - inputCore.InputManager.executeGesture (08:08:31.844):
Input: kb(desktop):upArrow
IO - speech.speak (08:08:31.864):
Speaking [LangChangeCommand ('en_GB'), u'eSpeak NG']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:08:31.865):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:08:34.068):
Input: kb(desktop):upArrow
IO - inputCore.InputManager.executeGesture (08:08:34.763):
Input: kb(desktop):downArrow
IO - speech.speak (08:08:34.778):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Speech API version 4']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:08:34.779):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:08:35.915):
Input: kb(desktop):upArrow
IO - speech.speak (08:08:35.934):
Speaking [LangChangeCommand ('en_GB'), u'eSpeak NG']
DEBUGWARNING - synthDrivers.sapi5.SynthDriver.speak (08:08:35.936):
Unsupported speech command: LangChangeCommand ('en_GB')
IO - inputCore.InputManager.executeGesture (08:08:36.915):
Input: kb(desktop):enter
INFO - synthDrivers.espeak.SynthDriver.__init__ (08:08:36.923):
Using eSpeak NG version 1.49.3 dev
DEBUG - speechDictHandler.SpeechDict.load (08:08:36.954):
Loading speech dictionary '.\userConfig\speechDicts\voiceDicts.v1\espeak\espeak-English (Great Britain).dic'...
DEBUG - speechDictHandler.SpeechDict.load (08:08:36.956):
9 loaded records.
DEBUG - synthDriverHandler.SynthDriver.loadSettings (08:08:36.956):
Loaded settings for SynthDriver espeak
INFO - synthDriverHandler.setSynth (08:08:36.956):
Loaded synthDriver espeak
DEBUG - gui.settingsDialogs.SettingsDialog._setInstanceDestroyedState (08:08:36.957):
Setting state to destroyed for instance: <gui.settingsDialogs.SynthesizerSelectionDialog object at 0x05C743A0>

Snip
going to voice or variants as it should be.

Current _instances {<gui.settingsDialogs.NVDASettingsDialog object at 0x05D84DA0>: 0, <gui.settingsDialogs.SynthesizerSelectionDialog object at 0x05C743A0>: 1}
IO - inputCore.InputManager.executeGesture (08:08:40.803):
Input: kb(desktop):tab
IO - speech.speak (08:08:40.821):
Speaking [LangChangeCommand ('en_GB'), u'Change... button Alt+h']
IO - inputCore.InputManager.executeGesture (08:08:41.789):
Input: kb(desktop):tab
IO - speech.speak (08:08:41.809):
Speaking [LangChangeCommand ('en_GB'), u'Voice: combo box Microsoft Server Speech Text to Speech Voice (en-GB, Hazel) collapsed Alt+v']
IO - inputCore.InputManager.executeGesture (08:08:47.099):
Input: kb(desktop):downArrow
IO - inputCore.InputManager.executeGesture (08:08:47.907):
Input: kb(desktop):upArrow
ERROR - unhandled exception (08:08:47.908):
Traceback (most recent call last):
File "gui\settingsDialogs.pyc", line 1012, in __call__
File "synthDriverHandler.pyc", line 33, in changeVoice
File "synthDrivers\sapi5.pyc", line 220, in _set_voice
File "synthDrivers\sapi5.pyc", line 165, in _getVoiceTokens
AttributeError: 'NoneType' object has no attribute 'getVoices'
IO - speech.speak (08:08:47.913):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Server Speech Text to Speech Voice (en-US, ZiraPro)']
IO - inputCore.InputManager.executeGesture (08:08:48.507):
Input: kb(desktop):upArrow
IO - inputCore.InputManager.executeGesture (08:08:49.227):
Input: kb(desktop):downArrow
ERROR - unhandled exception (08:08:49.229):
Traceback (most recent call last):
File "gui\settingsDialogs.pyc", line 1012, in __call__
File "synthDriverHandler.pyc", line 33, in changeVoice
File "synthDrivers\sapi5.pyc", line 220, in _set_voice
File "synthDrivers\sapi5.pyc", line 165, in _getVoiceTokens
AttributeError: 'NoneType' object has no attribute 'getVoices'
IO - speech.speak (08:08:49.237):
Speaking [LangChangeCommand ('en_GB'), u'Microsoft Server Speech Text to Speech Voice (en-GB, Hazel)']
IO - inputCore.InputManager.executeGesture (08:08:50.220):
Input: kb(desktop):tab
IO - speech.speak (08:08:50.229):
Speaking [LangChangeCommand ('en_GB'), u'Rate: slider 75 Alt+r']
IO - inputCore.InputManager.executeGesture (08:08:52.644):
Input: kb(desktop):tab
IO - speech.speak (08:08:52.661):
Speaking [LangChangeCommand ('en_GB'), u'Pitch: slider 50 Alt+p']
So the system is now totally confused, speaking with espeak the synth it thinks it has is sappi 5 but the voices of speech platform. No wonder some peculiar stuff is happening!
Looks to me that the variable of what is selected is not being stored or not being seen as things are changed within a session.
Brian



bglists@...
Sent via blueyonder.
Please address personal E-mail to:-
briang1@..., putting 'Brian Gaff'
in the display name field.
Newsgroup monitored: alt.comp.blind-users

----- Original Message -----
From: "Michael Curran" <mick@...>
To: <nvda-devel@groups.io>
Sent: Wednesday, June 05, 2019 9:58 AM
Subject: Re: [nvda-devel] speech refactor things in the threshold snapshot and introduction


Jamie had already ported sapi5 in his initial work in 2017.


However, I will look into this. Perhaps the characterMode command is now
somehow different to pre-speechRefactor.


Mick



On 6/5/2019 5:25 PM, Leonard de Ruijter wrote:
Hey Marco,

I was prett# sure that Mick also ported the sapi5 driver. I could be
wrong, though.

Regards,
Leonard

Op 5 jun. 2019 om 09:11 heeft Marco Zehe <@marcozehe
<mailto:@marcozehe>> het volgende geschreven:

Hi zvonimir,

this is totally expected. Threshold contains the refactored speech
engine. And it appears that backwards compatibility is not yet fully
restored. Neither the Vocalizer nor the SAPI 5 driver have been
updated to the new interfaces yet. The only synths that are expected
to work are eSpeak NG and Microsoft OneCore (Windows 10 voices).

Marco


Am Di., 4. Juni 2019 um 23:54 Uhr schrieb zvonimir stanečić, 9a5dsz
<zvonimirek222@... <mailto:zvonimirek222@...>>:

Hi to all,

My name is Zvonimir Stanecic, I am one of the NVDA translators
for Croatian language.

I have an issue with the latest threshold snaps, with the
codefactory vocalizer sapi5 synth, no matter which voice I use.

so… when trying to type individual letters, I hear the phoneme
names, but not the names of the letters, i.e characters.

For example, in Russian, I don’t hear hardsign and softsign/

Here is the sapi version of the codefactory vocalizer to download.

https://yadi.sk/d/kAenU1FJay2vxQ

It’s the same problem with the vocalizer for NVDA package from
codefactory.

Note that in alpha snaps this problem is not visible.

Best,

Zvonimir