1. Nvda | Nvda
  2. Topics

NVDA reading speed

Sarah k Alawami
 

I've listened to the vocalizer speed at about 100, slower with rate boost,
same with one core, I have to listen fast as I want to cram as much info in
as I can as fast as possible. I will slow down for demos, unless I'm
describing what I'm hearing. Then I will not.

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Chris Shook
Sent: Friday, December 30, 2022 7:24 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA reading speed

I know most people prefer the robitic voices, but I use Vocalizer
expressive.
What is a good speed for that synthesizer.
Brian's Mail list account
 

I was not blind to start with, and this may well reflect on the reading speed I use. The voice in Microsoft Edge seems pretty good to me for reading stuff, and I tend to just convert stuff to html or pdf to read in that. However I've not yet delved into anything more advanced than Daniel etc, for reading in NVDA. They are passable, but if you listen to the real John Briggs reading, its still clear we have a long way to go to give accurate inflection and expression to voices.
Brian

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

----- Original Message -----
From: "Chris Shook" <chris0309@...>
To: <nvda@nvda.groups.io>
Sent: Thursday, December 29, 2022 8:20 PM
Subject: Re: [nvda] NVDA reading speed


I keep mine slower than most. I think it's because I read using audio
books growing up and I like mine to be as close to that speed as possible.
Monty have you been blind long?
Brian Vogel
 

On Fri, Dec 30, 2022 at 12:52 PM, Brian's Mail list account wrote:
its still clear we have a long way to go to give accurate inflection and expression to voices.
-
I disagree.  The issue is not the ability of modern synthesizers to produce stunningly human-sounding voices, but the ability of the synthesizers (and/or what feeds them) to recognize the actual intent in a lot of text the way humans do.

There are scads of inflections for almost anything that can radically alter the meaning, but the technology that's "reading" or processing the text prior to it being sent for speech synthesis frequently does not recognize the intent behind the text.  This is another area where AI would enter the picture to clarify the intent of text prior to it being sent for speech synthesis.

The phrase, "Yeah, right," can mean two diametrically opposed things depending on inflection, so you're absolutely correct that unless the synth produces the correct inflection for the intent the meaning is not communicated in any way accurately.  That's far from the only instance.
--

Brian - Virginia, USA - Windows 10 Pro, 64-Bit, Version 22H2, Build 19045; Office 2016, Version 16.0.15726.20188, 32-bit

"Be Yourself" is the worst advice you can give to some people.

       ~ Tom Masson
Louise Pfau
 

Hi.  I'm used to having the screen reader speed set quite high now because that's how I had to have it in university so I could get through my course material.  When I was learning to use a screen reader, I would gradually increase the speed as I became more comfortable.
--
Louise
Windows 10 Pro 22H2 (OS Build 19045.2364).
NVDA 2022.3.2.
Gene
 

Some synthesizers of the newer variety are very bad at pausing properly.  Some may not be.  But I've heard enough of them that that is one reason I still use Via Voice which is very similar to Eloquence.  and I am using it legally, by the way.  I have it as a result of purchasing programs that use it and its use is unrestricted as far as other programs seeing an working with it is concerned.

Via Voice has good rules for determining pauses and inflection.  I'm not saying they are of human speaker quality but they are good enough.  If they weren't, I'd be the first to complain, being very sensitive to such things, having been active in a drama school during my teenage years.

Dectalk, largely unused these days, has good pausing rules as well.

Gene.

On 12/30/2022 11:59 AM, Brian Vogel wrote:

On Fri, Dec 30, 2022 at 12:52 PM, Brian's Mail list account wrote:
its still clear we have a long way to go to give accurate inflection and expression to voices.
-
I disagree.  The issue is not the ability of modern synthesizers to produce stunningly human-sounding voices, but the ability of the synthesizers (and/or what feeds them) to recognize the actual intent in a lot of text the way humans do.

There are scads of inflections for almost anything that can radically alter the meaning, but the technology that's "reading" or processing the text prior to it being sent for speech synthesis frequently does not recognize the intent behind the text.  This is another area where AI would enter the picture to clarify the intent of text prior to it being sent for speech synthesis.

The phrase, "Yeah, right," can mean two diametrically opposed things depending on inflection, so you're absolutely correct that unless the synth produces the correct inflection for the intent the meaning is not communicated in any way accurately.  That's far from the only instance.
--

Brian - Virginia, USA - Windows 10 Pro, 64-Bit, Version 22H2, Build 19045; Office 2016, Version 16.0.15726.20188, 32-bit

"Be Yourself" is the worst advice you can give to some people.

       ~ Tom Masson
Monte Single
 

And the perfect example of the synth knowing which way to say a word is the  word “read”.

Now, should that be pronounced reed or red?

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian Vogel
Sent: December 30, 2022 12:00 PM
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA reading speed

 

On Fri, Dec 30, 2022 at 12:52 PM, Brian's Mail list account wrote:

its still clear we have a long way to go to give accurate inflection and expression to voices.

-
I disagree.  The issue is not the ability of modern synthesizers to produce stunningly human-sounding voices, but the ability of the synthesizers (and/or what feeds them) to recognize the actual intent in a lot of text the way humans do.

There are scads of inflections for almost anything that can radically alter the meaning, but the technology that's "reading" or processing the text prior to it being sent for speech synthesis frequently does not recognize the intent behind the text.  This is another area where AI would enter the picture to clarify the intent of text prior to it being sent for speech synthesis.

The phrase, "Yeah, right," can mean two diametrically opposed things depending on inflection, so you're absolutely correct that unless the synth produces the correct inflection for the intent the meaning is not communicated in any way accurately.  That's far from the only instance.
--

Brian - Virginia, USA - Windows 10 Pro, 64-Bit, Version 22H2, Build 19045; Office 2016, Version 16.0.15726.20188, 32-bit

"Be Yourself" is the worst advice you can give to some people.

       ~ Tom Masson
Josh Kennedy
 

When I use espeak, I find that the faster I read I have to turn the inflection down because the default inflection of 75 is too distracting. It's almost as if the speech is trying to hard to sound too natural. So I usually put the inflection down to 20 or 30 when I'm reading very fast with. espeak, and also when I am just navigating around or I want to get through emails quickly I mostly let the inflection at 20 or 30 and I put the race the rate boost I turn that on and I usually have my speech rate set to around 20 or 25%

21 - 27 of 27