On Jan 30, 2023, at 7:38 AM, Josh Kennedy <joshknnd1982@...> wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

Hi,

Add-ons discovery: that's something NV Access is interested in through the concept of an add-on store.

As for bundling add-ons with NVDA: in theory, yes; in practice, no. Two top issues are security (add-ons can do various things as it will run with same privileges as NVDA) and licensing (not all ad-on components are licensed under GPL 2). I believe we had debates about this before, and we had mixed reception to the idea of bundling add-ons alongside NVDA (that was a while back).

Cheers,

Joseph

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true. 

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true. 

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster. 

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene

On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true. 

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

I think you confused a few terms in your message, but to be fair, I run the ivan3 voice at 99 percent, and it works just fine for me.  I like the ivan3 voice, because it's louder than most of the other voices, and although I did have to adjust the pitch somewhat, I have no trouble understanding it in any way.  I like it being a bit louder, because that means that (especially on a laptop), I can have the system volume lower, thus saving battery, because driving the amplifier inside the laptop does take power, and although it's probably not significant by itself, I've also found that removing programs from the tasklist that start when windows does, also helps to cut down on battery usage.  Those two things combined give me something like 10-15 minutes of additional battery usage, which may not seem significant, but it sure does help those times when sockets aren't readily available.  Plus, anything that reduces load on the cpu is a plus in my book.

I keep the speech to 99 percent, because I find that at 100 percent, it does indeed break up too much to be usable on an ongoing basis.  I find the same problem on the iPhone, those voices I have to run at 95 percent, because 100 percent breaks them up as well, (especially the enhanced voices), once they get above a certain percentage speed wise, they sound even worse than the nonenhanced versions of the same voice.  I always tell folks Samantha enhanced, above 90 percent or so sounds like she has a sore throat, so I've given up on the enhanced voices on apple devices, and just use the regular ones, because I can run them faster than the enhanced ones, and they still sound good to me.
I've used most of the voices provided with NVDA at one point or another, and I've found that almost all of them suffer from the breaking up of speech if the speed gets high enough.  Some voices start before the 100 percent mark, but those are relatively rare.  Generally I find that the speech is find up to the 99 percent mark, but beyond that not so much.  This is in direct opposition to the voices that come with the mac, on OSX, I was able to take advantage of a bug in apple's speed settings to crank the speech up to 150 or even 200 percent in some cases before it became unintelligible.  Not sure what the difference is, but that's been my experience.

Obviously, your mileage may vary.

On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster. 

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene

On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true. 

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster. 

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene

On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true. 

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

hi,

Last night I did something I thought impossible. A friend of mine is, and
was, a user of eloquence tts for many years. Then I showed him the changes I
made to espeak, asking him what tts I was using. He had to pause for a
second before answering, espeak. I then asked him if he liked the much
improved espeak. He sai that he did. After I helped him set the variant to
Reed, set inflection to 30, lower the pitch by a small amount, he was quite
happy with espeak. He also complained that the reason he still uses Jaws is
because jaws has a bunch of features NVDA does not have. I told him you can
find most of those features in NVDA addons. He complained that Jaws is nice
because it gives you everything in one package already there. so therefore
my suggestions are these.

1. Instead of the addons manager putting you onto a website and you have to
go find the addons, please make it more like how window-eyes did things. Add
a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links.
Put them in a list view with checkboxes. have their descriptions in a read
only edit box and an install button and uninstall buttons beside the edit
box. This will make it easier for new users to find and install the addons
they want, even installing multiple addons if they choose. You could even
filter addons by type of addon, if its a voice addon like vocalizer or an
addon that just adds new features you want? If its vocalizer it will let you
download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA
full packages. NVDA advanced or full would be much larger than the basic
NVDA because it would contain all compatible NVDA addons. And NVDA basic
would be the small traditional NVDA, the one with no addons by default. The
one we currently download. NVDA full would be great for new NVDA users
because it would just have everything they need or would want in the future
all addons preinstalled. They could hit a keystroke for helpful hints and
just use the features they want as they become familiar with them. NVDA full
would be great for assistive technology trainers and employers because
everything they would ever want and need is right there all addons
preinstalled. Want to relabel an object in your work application? The object
enhancer addon is already there.

Oh, and I wanted to comment on the dictionary corrections of voices.  I think you'd be surprised if you checked the dictionary entries, and/or the voice source.  I think you'll find there's way more word corrections than you'd otherwise think, regardless of the synth you're using.

Often times, there's built-in dictionaries that aren't accessible to the end user that change pronunciations without the end user even being aware those entries exist.  English is a complicated language, so this isn't really a surprise.  Add to that, the existence of several different dialects, and you run into the case where the developer of a particular synth thinks a word should be pronounced one way, and a user from another part of the country hears that as the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.

On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster. 

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene

On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true. 

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these. 

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it. 

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there. 

I haven't used the newer kind of synthesizers to any extent but Eloquence, compared with the small amount I've used some newer synthesizers, pronounces the most words correctly without correction.  I used Dectalk a lot before Eloquence and Eloquence is more accurate out of the box.  For whatever reason, that is my experience.
Gene
On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices. I think you'd be surprised if you checked the dictionary entries, and/or the voice source.  I think you'll find there's way more word corrections than you'd otherwise think, regardless of the synth you're using.

Often times, there's built-in dictionaries that aren't accessible to the end user that change pronunciations without the end user even being aware those entries exist.  English is a complicated language, so this isn't really a surprise.  Add to that, the existence of several different dialects, and you run into the case where the developer of a particular synth thinks a word should be pronounced one way, and a user from another part of the country hears that as the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there.

I have always thought that Eloquence was the leasst human sounding
speech syn so I stick with vocalizer.

On 1/30/2023 2:12 PM, Gene wrote:

I haven't used the newer kind of synthesizers to any extent but
Eloquence, compared with the small amount I've used some newer
synthesizers, pronounces the most words correctly without correction.  I
used Dectalk a lot before Eloquence and Eloquence is more accurate out
of the box.  For whatever reason, that is my experience.

Gene

On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices.
I think you'd be surprised if you checked the dictionary entries,
and/or the voice source.  I think you'll find there's way more word
corrections than you'd otherwise think, regardless of the synth you're
using.

Often times, there's built-in dictionaries that aren't accessible to
the end user that change pronunciations without the end user even
being aware those entries exist.  English is a complicated language,
so this isn't really a surprise.  Add to that, the existence of
several different dialects, and you run into the case where the
developer of a particular synth thinks a word should be pronounced one
way, and a user from another part of the country hears that as the
wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply
don't believe you can make E-Speak sound enough like NVDA and perform
enough like it at fast speech rates to be anywhere near satisfactory
to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with
Eloquence but is getting to be reasonably fast, speech is annoyingly
choppy and much harder to understand than Eloquence, even when
Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak
pronounces nearly as many words correctly, without dictionary
correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical
of your claim.  How can we try Reed and/or send a recording of
E-speak reading at the speed of 56 and perhaps a sample of it
speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all
representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't
have an opinion.  But I have never, and you have often said things
like, this synthesizer, or these changes, sound like Eloquence,
found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine
is, and was, a user of eloquence tts for many years. Then I showed
him the changes I made to espeak, asking him what tts I was using.
He had to pause for a second before answering, espeak. I then asked
him if he liked the much improved espeak. He sai that he did. After
I helped him set the variant to Reed, set inflection to 30, lower
the pitch by a small amount, he was quite happy with espeak. He
also complained that the reason he still uses Jaws is because jaws
has a bunch of features NVDA does not have. I told him you can find
most of those features in NVDA addons. He complained that Jaws is
nice because it gives you everything in one package already there.
so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and you
have to go find the addons, please make it more like how
window-eyes did things. Add a web scraper to the addons manager
that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download
links. Put them in a list view with checkboxes. have their
descriptions in a read only edit box and an install button and
uninstall buttons beside the edit box. This will make it easier for
new users to find and install the addons they want, even installing
multiple addons if they choose. You could even filter addons by
type of addon, if its a voice addon like vocalizer or an addon that
just adds new features you want? If its vocalizer it will let you
download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced
or NVDA full packages. NVDA advanced or full would be much larger
than the basic NVDA because it would contain all compatible NVDA
addons. And NVDA basic would be the small traditional NVDA, the one
with no addons by default. The one we currently download. NVDA full
would be great for new NVDA users because it would just have
everything they need or would want in the future all addons
preinstalled. They could hit a keystroke for helpful hints and just
use the features they want as they become familiar with them. NVDA
full would be great for assistive technology trainers and employers
because everything they would ever want and need is right there all
addons preinstalled. Want to relabel an object in your work
application? The object enhancer addon is already there.

I have always thought that Eloquence  was the leasst human sounding speech syn so I stick with vocalizer.

On 1/30/2023 2:12 PM, Gene wrote:

I haven't used the newer kind of synthesizers to any extent but Eloquence, compared with the small amount I've used some newer synthesizers, pronounces the most words correctly without correction.  I used Dectalk a lot before Eloquence and Eloquence is more accurate out of the box. For whatever reason, that is my experience.

Gene

On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices.  I think you'd be surprised if you checked the dictionary entries, and/or the voice source.  I think you'll find there's way more word corrections than you'd otherwise think, regardless of the synth you're using.

Often times, there's built-in dictionaries that aren't accessible to the end user that change pronunciations without the end user even being aware those entries exist.  English is a complicated language, so this isn't really a surprise.  Add to that, the existence of several different dialects, and you run into the case where the developer of a particular synth thinks a word should be pronounced one way, and a user from another part of the country hears that as the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there.

.

I haven't used the newer kind of synthesizers to any extent but
Eloquence, compared with the small amount I've used some newer
synthesizers, pronounces the most words correctly without correction.
I used Dectalk a lot before Eloquence and Eloquence is more accurate
out of the box. For whatever reason, that is my experience.

Gene

On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices.
I think you'd be surprised if you checked the dictionary entries,
and/or the voice source. I think you'll find there's way more word
corrections than you'd otherwise think, regardless of the synth
you're using.

Often times, there's built-in dictionaries that aren't accessible to
the end user that change pronunciations without the end user even
being aware those entries exist. English is a complicated language,
so this isn't really a surprise. Add to that, the existence of
several different dialects, and you run into the case where the
developer of a particular synth thinks a word should be pronounced
one way, and a user from another part of the country hears that as
the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply
don't believe you can make E-Speak sound enough like NVDA and
perform enough like it at fast speech rates to be anywhere near
satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with
Eloquence but is getting to be reasonably fast, speech is annoyingly
choppy and much harder to understand than Eloquence, even when
Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak
pronounces nearly as many words correctly, without dictionary
correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very
skeptical of your claim. How can we try Reed and/or send a
recording of E-speak reading at the speed of 56 and perhaps a sample
of it speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all
representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't
have an opinion. But I have never, and you have often said things
like, this synthesizer, or these changes, sound like Eloquence,
found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine
is, and was, a user of eloquence tts for many years. Then I showed
him the changes I made to espeak, asking him what tts I was using.
He had to pause for a second before answering, espeak. I then
asked him if he liked the much improved espeak. He sai that he
did. After I helped him set the variant to Reed, set inflection to
30, lower the pitch by a small amount, he was quite happy with
espeak. He also complained that the reason he still uses Jaws is
because jaws has a bunch of features NVDA does not have. I told
him you can find most of those features in NVDA addons. He
complained that Jaws is nice because it gives you everything in one package already there.
so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and
you have to go find the addons, please make it more like how
window-eyes did things. Add a web scraper to the addons manager
that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download
links. Put them in a list view with checkboxes. have their
descriptions in a read only edit box and an install button and
uninstall buttons beside the edit box. This will make it easier
for new users to find and install the addons they want, even
installing multiple addons if they choose. You could even filter
addons by type of addon, if its a voice addon like vocalizer or an
addon that just adds new features you want? If its vocalizer it
will let you download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced
or NVDA full packages. NVDA advanced or full would be much larger
than the basic NVDA because it would contain all compatible NVDA
addons. And NVDA basic would be the small traditional NVDA, the
one with no addons by default. The one we currently download. NVDA
full would be great for new NVDA users because it would just have
everything they need or would want in the future all addons
preinstalled. They could hit a keystroke for helpful hints and
just use the features they want as they become familiar with them.
NVDA full would be great for assistive technology trainers and
employers because everything they would ever want and need is
right there all addons preinstalled. Want to relabel an object in
your work application? The object enhancer addon is already there.

Not trying to be overly harsh in what I'm about to say, but I honestly don't understand how anybody is able to use E-Speak on a regular basis. At best, to my admittedly subjective ears, the voices are metallic and unnatural sounding, and at worst, are almost unintelligible. The developers of E-Speak, who for all I know may have the best of intentions, have packed it full of what I consider novelty or joke voices, ala Mr. Serious, Anxious Andy, whatever that "announcement system" voice is, robot voices in space, etc. With Eloquence, the Microsoft voices, etc. most of the voices will work right out of the box, once I've set the speed to my liking, but with E-speak, I'd have to practically perform surgery on the voices, setting various controls in an effort to try to make it somewhat listenable. Needless to say, I won't be converted from Eloquence any time soon.

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Don H
Sent: Monday, January 30, 2023 2:18 PM
To: nvda@nvda.groups.io
Subject: Re: [nvda] converting a long time eloquence user

I have always thought that Eloquence was the leasst human sounding speech syn so I stick with vocalizer.

On 1/30/2023 2:12 PM, Gene wrote:

I haven't used the newer kind of synthesizers to any extent but
Eloquence, compared with the small amount I've used some newer
synthesizers, pronounces the most words correctly without correction.
I used Dectalk a lot before Eloquence and Eloquence is more accurate
out of the box. For whatever reason, that is my experience.

Gene

On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices.
I think you'd be surprised if you checked the dictionary entries,
and/or the voice source. I think you'll find there's way more word
corrections than you'd otherwise think, regardless of the synth
you're using.

Often times, there's built-in dictionaries that aren't accessible to
the end user that change pronunciations without the end user even
being aware those entries exist. English is a complicated language,
so this isn't really a surprise. Add to that, the existence of
several different dialects, and you run into the case where the
developer of a particular synth thinks a word should be pronounced
one way, and a user from another part of the country hears that as
the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply
don't believe you can make E-Speak sound enough like NVDA and
perform enough like it at fast speech rates to be anywhere near
satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with
Eloquence but is getting to be reasonably fast, speech is annoyingly
choppy and much harder to understand than Eloquence, even when
Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak
pronounces nearly as many words correctly, without dictionary
correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very
skeptical of your claim. How can we try Reed and/or send a
recording of E-speak reading at the speed of 56 and perhaps a sample
of it speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all
representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't
have an opinion. But I have never, and you have often said things
like, this synthesizer, or these changes, sound like Eloquence,
found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine
is, and was, a user of eloquence tts for many years. Then I showed
him the changes I made to espeak, asking him what tts I was using.
He had to pause for a second before answering, espeak. I then
asked him if he liked the much improved espeak. He sai that he
did. After I helped him set the variant to Reed, set inflection to
30, lower the pitch by a small amount, he was quite happy with
espeak. He also complained that the reason he still uses Jaws is
because jaws has a bunch of features NVDA does not have. I told
him you can find most of those features in NVDA addons. He
complained that Jaws is nice because it gives you everything in one package already there.
so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and
you have to go find the addons, please make it more like how
window-eyes did things. Add a web scraper to the addons manager
that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download
links. Put them in a list view with checkboxes. have their
descriptions in a read only edit box and an install button and
uninstall buttons beside the edit box. This will make it easier
for new users to find and install the addons they want, even
installing multiple addons if they choose. You could even filter
addons by type of addon, if its a voice addon like vocalizer or an
addon that just adds new features you want? If its vocalizer it
will let you download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced
or NVDA full packages. NVDA advanced or full would be much larger
than the basic NVDA because it would contain all compatible NVDA
addons. And NVDA basic would be the small traditional NVDA, the
one with no addons by default. The one we currently download. NVDA
full would be great for new NVDA users because it would just have
everything they need or would want in the future all addons
preinstalled. They could hit a keystroke for helpful hints and
just use the features they want as they become familiar with them.
NVDA full would be great for assistive technology trainers and
employers because everything they would ever want and need is
right there all addons preinstalled. Want to relabel an object in
your work application? The object enhancer addon is already there.

Not trying to be overly harsh in what I'm about to say, but I honestly don't understand how anybody is able to use E-Speak on a regular basis. At best, to my admittedly subjective ears, the voices are metallic and unnatural sounding, and at worst, are almost unintelligible. The developers of E-Speak, who for all I know may have the best of intentions, have packed it full of what I consider novelty or joke voices, ala Mr. Serious, Anxious Andy, whatever that "announcement system" voice is, robot voices in space, etc. With Eloquence, the Microsoft voices, etc. most of the voices will work right out of the box, once I've set the speed to my liking, but with E-speak, I'd have to practically perform surgery on the voices, setting various controls in an effort to try to make it somewhat listenable. Needless to say, I won't be converted from Eloquence any time soon.

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Don H
Sent: Monday, January 30, 2023 2:18 PM
To: nvda@nvda.groups.io
Subject: Re: [nvda] converting a long time eloquence user

I have always thought that Eloquence was the leasst human sounding speech syn so I stick with vocalizer.

On 1/30/2023 2:12 PM, Gene wrote:

I haven't used the newer kind of synthesizers to any extent but
Eloquence, compared with the small amount I've used some newer
synthesizers, pronounces the most words correctly without correction.
I used Dectalk a lot before Eloquence and Eloquence is more accurate
out of the box. For whatever reason, that is my experience.

Gene

On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices.
I think you'd be surprised if you checked the dictionary entries,
and/or the voice source. I think you'll find there's way more word
corrections than you'd otherwise think, regardless of the synth
you're using.

Often times, there's built-in dictionaries that aren't accessible to
the end user that change pronunciations without the end user even
being aware those entries exist. English is a complicated language,
so this isn't really a surprise. Add to that, the existence of
several different dialects, and you run into the case where the
developer of a particular synth thinks a word should be pronounced
one way, and a user from another part of the country hears that as
the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply
don't believe you can make E-Speak sound enough like NVDA and
perform enough like it at fast speech rates to be anywhere near
satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen
with Eloquence but is getting to be reasonably fast, speech is
annoyingly choppy and much harder to understand than Eloquence,
even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak
pronounces nearly as many words correctly, without dictionary
correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very
skeptical of your claim. How can we try Reed and/or send a
recording of E-speak reading at the speed of 56 and perhaps a
sample of it speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all
representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't
have an opinion. But I have never, and you have often said things
like, this synthesizer, or these changes, sound like Eloquence,
found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine
is, and was, a user of eloquence tts for many years. Then I
showed him the changes I made to espeak, asking him what tts I was using.
He had to pause for a second before answering, espeak. I then
asked him if he liked the much improved espeak. He sai that he
did. After I helped him set the variant to Reed, set inflection
to 30, lower the pitch by a small amount, he was quite happy with
espeak. He also complained that the reason he still uses Jaws is
because jaws has a bunch of features NVDA does not have. I told
him you can find most of those features in NVDA addons. He
complained that Jaws is nice because it gives you everything in one package already there.
so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and
you have to go find the addons, please make it more like how
window-eyes did things. Add a web scraper to the addons manager
that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and
download links. Put them in a list view with checkboxes. have
their descriptions in a read only edit box and an install button
and uninstall buttons beside the edit box. This will make it
easier for new users to find and install the addons they want,
even installing multiple addons if they choose. You could even
filter addons by type of addon, if its a voice addon like
vocalizer or an addon that just adds new features you want? If
its vocalizer it will let you download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced
or NVDA full packages. NVDA advanced or full would be much larger
than the basic NVDA because it would contain all compatible NVDA
addons. And NVDA basic would be the small traditional NVDA, the
one with no addons by default. The one we currently download.
NVDA full would be great for new NVDA users because it would just
have everything they need or would want in the future all addons
preinstalled. They could hit a keystroke for helpful hints and
just use the features they want as they become familiar with them.
NVDA full would be great for assistive technology trainers and
employers because everything they would ever want and need is
right there all addons preinstalled. Want to relabel an object in
your work application? The object enhancer addon is already there.

I have always thought that Eloquence  was the leasst human sounding speech syn so I stick with vocalizer.

On 1/30/2023 2:12 PM, Gene wrote:

I haven't used the newer kind of synthesizers to any extent but Eloquence, compared with the small amount I've used some newer synthesizers, pronounces the most words correctly without correction.  I used Dectalk a lot before Eloquence and Eloquence is more accurate out of the box. For whatever reason, that is my experience.

Gene

On 1/30/2023 1:49 PM, Travis Siegel wrote:

Oh, and I wanted to comment on the dictionary corrections of voices.  I think you'd be surprised if you checked the dictionary entries, and/or the voice source.  I think you'll find there's way more word corrections than you'd otherwise think, regardless of the synth you're using.

Often times, there's built-in dictionaries that aren't accessible to the end user that change pronunciations without the end user even being aware those entries exist.  English is a complicated language, so this isn't really a surprise.  Add to that, the existence of several different dialects, and you run into the case where the developer of a particular synth thinks a word should be pronounced one way, and a user from another part of the country hears that as the wrong pronunciation, so that complicates things too.

Never mind accents from different parts of the world.


On 1/30/2023 12:47 PM, Gene wrote:

If you tell me how to get and use Reed, I may try it, but I simply don't believe you can make E-Speak sound enough like NVDA and perform enough like it at fast speech rates to be anywhere near satisfactory to someone who wants NVDA.

When I set the speech rate to 56, which is slower than I listen with Eloquence but is getting to be reasonably fast, speech is annoyingly choppy and much harder to understand than Eloquence, even when Eloquence is set to read at much faster rates.

I haven't tested this, but I would be very surprised if E-Speak pronounces nearly as many words correctly, without dictionary correction, as Eloquence.

Unless you have significantly redesigned E-speak, I am very skeptical of your claim.  How can we try Reed and/or send a recording of E-speak reading at the speed of 56 and perhaps a sample of it speaking somewhat faster.

You may have persuaded one listener, but one listener isn't at all representative of how others may react.

Gene
On 1/30/2023 11:28 AM, Gene via groups.io wrote:

I haven't given any thought to what you are suggesting and I don't have an opinion.  But I have never, and you have often said things like, this synthesizer, or these changes, sound like Eloquence, found any statements from anyone of the sort to be true.

Gene

On 1/30/2023 9:38 AM, Josh Kennedy wrote:

hi,

Last night I did something I thought impossible. A friend of mine is, and was, a user of eloquence tts for many years. Then I showed him the changes I made to espeak, asking him what tts I was using. He had to pause for a second before answering, espeak. I then asked him if he liked the much improved espeak. He sai that he did. After I helped him set the variant to Reed, set inflection to 30, lower the pitch by a small amount, he was quite happy with espeak. He also complained that the reason he still uses Jaws is because jaws has a bunch of features NVDA does not have. I told him you can find most of those features in NVDA addons. He complained that Jaws is nice because it gives you everything in one package already there. so therefore my suggestions are these.

1. Instead of the addons manager putting you onto a website and you have to go find the addons, please make it more like how window-eyes did things. Add a web scraper to the addons manager that will go out to both:

https://addons.nvda-project.org/index.en.html

and also:

https://nvda-addons.org/
and

https://github.com/BabbageCom/objEnhancer

scrape those sites for all addons, their descriptions and download links. Put them in a list view with checkboxes. have their descriptions in a read only edit box and an install button and uninstall buttons beside the edit box. This will make it easier for new users to find and install the addons they want, even installing multiple addons if they choose. You could even filter addons by type of addon, if its a voice addon like vocalizer or an addon that just adds new features you want? If its vocalizer it will let you download the addon but its still up to you to purchase it.

2. Offer two NVDA download options, NVDA basic, and NVDA advanced or NVDA full packages. NVDA advanced or full would be much larger than the basic NVDA because it would contain all compatible NVDA addons. And NVDA basic would be the small traditional NVDA, the one with no addons by default. The one we currently download. NVDA full would be great for new NVDA users because it would just have everything they need or would want in the future all addons preinstalled. They could hit a keystroke for helpful hints and just use the features they want as they become familiar with them. NVDA full would be great for assistive technology trainers and employers because everything they would ever want and need is right there all addons preinstalled. Want to relabel an object in your work application? The object enhancer addon is already there.