2017-06-29 16:46:16

Wait, Colton you got the vocalizer expressive 2.2 voices like Nathan working with NVDA? How...

2017-06-29 18:49:46

Is what you're asking for even possible? It looks as though the format has changed completely. Or maybe that's just specific to iOS and macOS.

Fun fact: jailbreakers had the Siri voices in VoiceOver before Apple introduced the feature. This indicates, quite clearly, that jailbreakers provide valuable direction for future development.

Just myself, as usual.

2017-06-30 00:03:51

I would like to clarify some Points which might bring confusion if not properly explained.

1. I did not get Vocalizer Expressive v2.1.0 to work with NVDA.
Yes, CodeFactory did, but I did not get, how they did it.
Point is that usually you have one Folder per language, not per voice.
When you look at Vocalizer 2 for JAWS, this rule still is in place, sadly the Commercial NVDA Driver doesn't care, if I remember right.

2. The nuan_platform.dll file found in the Driver from http://www.vocalizer-nvda.com seems to be incapable of cooping with Versions of ve.dll higher than v1.3.


3. You can get v1.3 voice files and the required DLL from any official Trial Versions of Dolphin Super Nova.

4. Vocalizer Expressive V1 for JAWS uses an older ve.dll, namely v1.1 and has less voices for JAWS 17 and older.


5. The voices offered on http://www.vocalizer-nvda.com are presumably v1.1 voices, but can be used with v1.3.
However, newer Versions of certain voices have more extra files, which the Vocalizer for NVDA product is excluding.

6. The Vocalizer For NVDA product has to my knowledge two voices which are different than the rest.
In their case These are the German voice Petra and the French voice Audrey. These Versions are bigger than their 1.3 variants and they are named differently.
According to several incomplete datasheets, there are supposed to be voices which can speak certain words of other languages directly.
This is supposed to mean that for example you could have a German voice read a German text with a sentence in English.
In theory the voice could speak that sentence in proper English and return to read German afterwards without having to perform a voice Switch for the part in a different language.

However since Tiflotecnia, Lda built a simplistic Driver, this secondary language speaking Feature is impossible, especially if you are letting NVDA do the reading in MS Word for example.

I am assuming that a proper Vocalizer Distribution would have ist own program compared to TextAloud, which would let the user read text and possibly modify voice Parameters with Special tags, which obviously would not work if NVDA passes text to the TTS engine via a third rate custom made Interface...

Finally, I want to say, that it is possible to use a

2017-06-30 00:51:23

you can also find them more easily at www.***

2017-06-30 01:47:17 (edited by st.mc 2017-06-30 01:48:29)

@jack yes! It's possible to have vocalizer 2.2 voices on nvda. both with SAPI and the NVDA addon. If you havn't got your hands on that yet, Just send me a PM. I'll be more than happy to help.
Check blindhelp out by the way, It's a good source.

:)

2017-06-30 03:59:56 (edited by jack 2017-06-30 04:00:19)

Yeah, so I heard. I actually got it from blindhelp, if that's the package you're referring to. Currently using Nathan and he works flawlessly.

2017-06-30 16:06:01

Is this new package backwards compatible, meaning if it is possible to bring the Siri voices into this, or do you have to run multiple Versions like one for v1.3 and one for v2.2?
And I also would like to know if the nuan_platform.dll was modified to accept the ve.dll with Version 2.2?

Is it also possible to use the v1.5 voices which were available for Android before they updated to v2.2?

Furthermore, how do you set this v2.2 engine up for SAPI5?
Do the different Versions of ve.dll require different registry values and class ids?

And finally, arere there v2.x Versions of the Siri voices, or can you run them only with Version 1.3, if so, why don't the British English voices work with a fully functional v1.3 package?

2017-06-30 16:37:31

I don't think the voices are backwords compatible. What I did was keep the 1.3 addon around, and disable it with nvda so that one addon can be running at a time. Unlike blind extra, the 2.2 addon is no parlor trick, it is, true to its word, a modified version of the tiflotechnia driver that has the 2.2 dll's. That's probably also why only one can be running at a time. All the 2.2 voices are available on the site. Unfortunately the only compact bundle they have is all languages, and I'm half tempted to, if I can, download that, and then strip out all the voices of the languages that i don't need. As for the Siri voices I highly doubt the 1.3 ones will work in 2.2. So the 2.2 ones will have to be ripped out of either ios or the mac.

2017-06-30 19:31:37

I still don't get why no official product gets updated.
For example, Window Eyes has the 1.3 voices, which for example includes the German voices Anna and Markus (there are other voices to put in that spot), but JAWS did Keep it at v1.1 and did nothing until they put out that now v2.2 is available.
But if you looked at what is on the JAWS page, for another example, the JAWS 18 voices miss Japanese entirely.

And then there is the Problem with the the voices officially on www.vocalizer-nvda.com.
As I stated before, These are 1.1 voices, but some are bigger than their regular Counterparts and are named like These: Petra-ml and Audrey-ml.

Supposedly this is not for fun but has an actual function, but neither vocalizer Driver can Access These extra Features.

No one knows whether v2 still has multilanguage voices with differently named files.

And this doesn't tell us why the so called 1.3 British Siri voices won't run properly with a working 1.3 Driver.

And I will put out a warning for all you here.
Do not atempt to use the CodeFactory voice modules in order to atempt to complete what *** is currently missing.
This atempt will cause a Crash or at least cause malfunctions, depending on what is installed and what is not.

You should know that CodeFactory is selling ist own Vocalizer and Eloquence package for NVDA, an English exclusive SAPI engine via Computer Room Services and the Vocalizer for Android app.

I can confirm that all These products use the same voice database.
According to Codefactory, Vocalizer 2 has at least four confirmed voice models: embedded-compact, embedded-pro, embedded-high and premium high.

Currently it is not possible to atempt to include missing variants or voices because for some reason the configuration files for all their voices with the Extension .hdr are not the same compared to other Vocalizer 2 distributions.

These .hdr files are supposed to be modified XML files and should thus be readable within any proper text Editor.
Theirs, however are not.

Thus I don't know how their Windows products actually work at all.

2017-07-01 08:55:29

here's an other thing. I'm wondering if it's possible to get these voices to recognize emojis. Yes, I know about the workaround with nvda, but how about sapi versions and vocalizer x on android?

:)

2017-07-01 16:39:15

St.mc, is there a vocalizer 2.2 distribution for android as well?

2017-07-01 21:35:32

I am not entirely sure.
I know that CodeFactory started selling Vocalizer Expressive and Eloquence together as one NVDA addon.
Furthermore, they sell SAPI Vocalizer Expressive and SAPI Eloquence.

As far as I know SAPI Vocalizer is sold through a British Company known as Computer Room Services.
However this SAPI vocalizer product only offers English voices with apparently not even the Option of buying other voices separately.
And before you ask Jack, originally they had Version 1.5 of Vocalizer Expressive for both SAPI and NVDA products before updating to Version 2.1 or 2.2 at the end of last year.

I know that CodeFactory is also selling Vocalizer for Android and I assume that this app got updated Content data at some Point.

If you want to look at These products, I can upload the official Trial Version installers.
They old Version should still work so that one product will download Version 1.5 and the other will install 2.x voices.
Unfortunately, you can't have both, since the CodeFactory product is one application and installing the new Version over the existing is handled like an update.
I also think that the update might derigister the 1.5 files but is not designed to remove or replace the files themselves from your Computer.
So installing the packages over each other is not recommended.

2017-07-01 23:41:09

Actually not. I tried vocalizer expressive on my braillenote a few months ago, when it was said that the pc versions got updated. No Nathan. The blindhelp site is pretty unclear, but the vocalizer expressive package they have for android said vocalizer e2. Maybe they will work with 2.2 voices.

2017-07-02 01:00:30

Or they made their own Version.
But I might be wrong there.

What I don't get is why no one is able to sell one definitive Vocalizer package with all voices in all variants from low Quality to high, with the official dictionary Editor, with the extra modules for reading foreign words, names of streets and Locations (for all the Navigation devices for cars) and so on.
And then there is supposedly a program in the SDK similar to Loquendo's Voice Experience Classic.

If you don't know what this is, it was a program similar to TextAloud, but made by Nuance in which you can read text, pre-process it and also insert tags for reading pre-recorded sentences like greetings for example.
And you are supposed to also have (at least for some voices) the same Feature Loquendo TTS had, namely speech tags for reading text more emotionally, e.g. sad, angry, or as a whisper.
And then there are supposed to be the pre-recorded things like laughs or different crying sounds you could place in a text file in a predetermined code which then would be interpreted and not read as-is.

But how are you supposed to use the Vocalizer dictionary in NVDA?
You can use dictionary Features in NVDA or JAWS.
But you can't share them between products.
If you could buy a proper Vocalizer product in case of Windows SAPI compatible, with all the required extra components, then you would Need only one dictionary for the rules you want  to create.
And Nuance make it clear in their product Information, that any Vocalizer from Automotive 5 to V6, to Expressive 1 and 2 have a user dictionary Editor, which strangely is missing from any known product.

And then there are the Siri voices.
If These are now longer Siri and IOS exclusive and can be System voices on IOS or OSX, then why the hell can't I get them offically on Windows?
And are you Aware that the offered Siri voices in this Topic (v1.3) are only in the Quality "Premium" and not "Premium High"?
Apparently no one knows if These voices exist in the highest Quality as well.
In theory they should, because all other Mainstream voices exist in five variants.
And there are more varriants than those five vor V1.3.
Supposedly for very limited devices you should also get them in 11 or 16 KHZ variants.
And then there is the language file.
Most voices use a language file like this: clc_ged_cfg3.dat.

The "ged" part in this example is for the German language definiton or whatever it does.
Most voices for v1.3 use a "cfg3.dat" file.

However someone sent me files taken from a device where a jailbreak had been applied to some Version of IOS 7.
Since it was a German phone I got two variants of the German voice Anna.
The compact voice and the Premium voice, not Premium High.
The compact Anna voice used a file named clc_ged_cfg1.dat.
This file is a lot smaller than the usual cfg3 file.
And the German voice Petra-ml (found only on www.vocalizer-nvda.com) uses this file: clc_ged_cfg4.dat.
This file is larger than the normal one.
Other languages also have cfg 4 files.
I think Ava or one of the others had it, allthough not every voice using a cfg4 file has the "-ml" Addition to ist Name.
You can install a -ml variant next to a normal variant and they don't block each other.

Sadly this is only Information for Version 1.3 and must not be confused with Version 2.2.

2017-07-02 02:51:06

The Vocalizer voices from www.harposoftware.com support SAPI 5 and the expressive tags. I haven't really looked at the Code Factory voices for Windows because their prices are too high. I purchased Ava from Harpo for $45 and am happy. The voices on Android aren't even premium quality. You get compact and plus. I don't know why Code Factory won't update the app to include at least premium quality. As for Apple, I'm guessing they use premium voices on iOS to save storage space. I'm pretty sure the Siri voices were exclusively developed for Apple, so they won't be available in the general Vocalizer product line.

The Siri voices do come in a premium high quality. I found this out when using Siri in macOS. I've also heard that some iOS 10 installations feature higher quality voices, although these don't appear to be available for VoiceOver and don't seem to be on all devices. I don't have these really enhanced Siri voices on my 16 GB iPhone 5S. As far as I know, there's no way to get voice data from macOS as the files are encrypted.

Grab my Adventure at C: stages Right here.

2017-07-02 08:08:41

@jack try this:
http://www.tyflonet.com/siciliano/VocalizerEx2_TTS.zip
this works with v2.2 voices like Nathan and Malcolm

:)

2017-07-02 08:52:24

so guys guess what!
I copied all the voice data that the nvda add-on needed to work from a Harpo software product, and I found some cool things.
When it reaches some specific words, it plays a pre-recorded sound. I guess it's those voice tags that you're talking about.
here are some examples:
Laughing!
Witching!
Whistleing!
Tuttuting!
Snoreing!
Raspberrying!
Hiccupsing!
Coughing!
Crying!
these tags won't work with all the voices. For instance, the examples given above only work with vocalizer Kate, and Allison's are different.

:)

2017-07-02 15:10:34

Michael, regarding the vocalizer 2.2 driver, the folks at blindhelp are definitely either making their own custom aftermarket distribution of the driver, or are finding it from some other source. I don't really know this team but whatever they're doing, they're certainly doing it right. As for Vocalizer Expresisve 2.2, wonderful. I'll uninstall 1.3 in that case since if it's anything like the nvda one, you can't have 2 versions at once.

2017-07-02 15:29:10

Correction Jack, you can have two Versions of Vocalizer Expressive.
At least for NVDA.
The *** Version Labels itself v2 and thus is a separate addon from the Standard Driver found on www.vocalizer-nvda.com.
You only Need to make sure that voices or voice packs are properly named, in the case of NVDA, you only Need to edit the mainefest.ini in the Archive.
In case you don't know, you can practically make anything into an NVDA addon by compressing it as a ZIP file with Standard compression.
The ZIP will have changed from .zip to .nvda-addon obviously.
It can have subfolders, but you must have a manifest.ini in the main Folder of the Archive.

I can currently run the German Siri voices in Version 1 while V2 is also installed.
Obviously I can't run both at the same time, but in the speech engine list in NVDA, I have two Vocalizer entries listed and can Switch at will.

There is more I could tell you, but I think that  it would not be appropriate for this talk to continue into that direction here in the Forum.

2017-07-02 15:31:04

I got the vocalizer voices package from blindHelp, when I installed it into NVDA, I have no vocalizer expressive on my synthesizer's lists. What do I need to do to fix this problem?

2017-07-02 18:15:38

Make sure you install all the visual c++ distributions. You *need* all of them, you can't just get the latest version.

2017-07-02 21:00:27

lord raven, pm me the one you've got for sapi. I want that. And if anybody has a working 1.3/1.5 with siri voices that actually, like, work, I'd like that. I only need siri obviously since everything else is ported. And for those who have the english compact bundle, which if you don't know this you can rip voices from C:\ProgramData\freedom scientific\vocalizer expressive\2.2\languages, try susan compact. Iww! That's how I got all the voices I wanted anyways, I didn't even know about blindhelp's packages. Somebody on skype leaked me a driver saying tiflotechnia version 4.0.0 alpha 2 or something.

----------
An anomaly in the matrix. An error in existence. A being who cannot get inside the goddamn box! A.K.A. Me.

2017-07-02 21:49:59

I'm not sure about the driver in blindhelp website, but I use this one,:
https://nvda.ru/uploads/vocalizer-drive … nvda-addon

:)

2017-07-03 05:02:32

@Jack, I have all the visual c++ distribution installed, still, I cannot see the voices.

2017-07-03 14:22:21

Download voices if you haven't already, and disable the 1.3. AT least that's what I did just to be safe.