NVDA's built-in OCR Capabilities


Richard Kuzma
 

Does anyone know if nvda has a function like the convenient ocr option in jaws where you can right click – do the applications key on a pdf and arrow down to convenient ocr with jaws and press enter and jaws will try and ocr the file and pull the text out of it?

Thanks,

Rich

 


Chris Smart
 

NVDA does have a feature called Windows OCR, which you can read all about by going to section 10.1 of the User Guide.

It won’t process a whole document though.

 

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Richard Kuzma via groups.io
Sent: September 8, 2022 8:50 AM
To: nvda@nvda.groups.io
Subject: [nvda] question on operation of nvda

 

Does anyone know if nvda has a function like the convenient ocr option in jaws where you can right click – do the applications key on a pdf and arrow down to convenient ocr with jaws and press enter and jaws will try and ocr the file and pull the text out of it?

Thanks,

Rich

 


Brian's Mail list account
 

Maybe I just have short documents then.
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 Smart" <ve3rwj@...>
To: <nvda@nvda.groups.io>
Sent: Thursday, September 08, 2022 2:05 PM
Subject: Re: [nvda] question on operation of nvda


NVDA does have a feature called Windows OCR, which you can read all about by
going to section 10.1 of the User Guide.

It won't process a whole document though.





From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Richard Kuzma
via groups.io
Sent: September 8, 2022 8:50 AM
To: nvda@nvda.groups.io
Subject: [nvda] question on operation of nvda



Does anyone know if nvda has a function like the convenient ocr option in
jaws where you can right click - do the applications key on a pdf and arrow
down to convenient ocr with jaws and press enter and jaws will try and ocr
the file and pull the text out of it?

Thanks,

Rich


Sarah k Alawami
 

No, however there is a nice add on called NAO, you might be able to find it on the community Spanish page. I use it now to scan pdfs and it’s nice.  It’s not as fancy as docuscan, however it worked for me when I needed to read a manual someone gave me .

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Richard Kuzma via groups.io
Sent: Thursday, September 8, 2022 5:50 AM
To: nvda@nvda.groups.io
Subject: [nvda] question on operation of nvda

 

Does anyone know if nvda has a function like the convenient ocr option in jaws where you can right click – do the applications key on a pdf and arrow down to convenient ocr with jaws and press enter and jaws will try and ocr the file and pull the text out of it?

Thanks,

Rich

 


Steve Nutt
 

No, without add-ons, NVDA will only scan the current screen.

JAWS will scan a whole document, whether it's imported from a PDF, or
acquired from a scanner directly.

Note: NVDA does have add-ons for this, but natively it can only scan the
current screen, as far as I know.

All the best

Steve

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian's Mail
list account via groups.io
Sent: 08 September 2022 15:20
To: nvda@nvda.groups.io
Subject: Re: [nvda] question on operation of nvda

Maybe I just have short documents then.
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 Smart" <ve3rwj@...>
To: <nvda@nvda.groups.io>
Sent: Thursday, September 08, 2022 2:05 PM
Subject: Re: [nvda] question on operation of nvda


NVDA does have a feature called Windows OCR, which you can read all about by
going to section 10.1 of the User Guide.

It won't process a whole document though.





From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Richard Kuzma
via groups.io
Sent: September 8, 2022 8:50 AM
To: nvda@nvda.groups.io
Subject: [nvda] question on operation of nvda



Does anyone know if nvda has a function like the convenient ocr option in
jaws where you can right click - do the applications key on a pdf and arrow
down to convenient ocr with jaws and press enter and jaws will try and ocr
the file and pull the text out of it?

Thanks,

Rich


 

Topic retitled for clarity and archival purposes.

But I also have a question:  Do any of the "internal" OCR options result in the file being OCRed having those changes kept permanently?  I can't imagine that the single page "quick and dirty" OCR would do this as it would make no sense to even try because many documents are more than one page, and a partially OCRed file is as bad as one that still needs to be processed.  But I don't know much about the NVDA Advanced OCR (NAO) Add-On other than that it exists and what its primary purpose is.

If the file is to be used over and over and over again, and neither adds a text layer that sticks with the file, there are other options that do just that.  But they're not done by NVDA itself or an add-on.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


Howard Traxler
 

Wondering about these OCR options:  Do all the offerred programs use the same OCR engine with just a little diferene in the interface?  Can't imagine that everyone wants to invent his own version of the wheel, as they say.


Howard

On 9/8/2022 10:33 AM, Brian Vogel wrote:
Topic retitled for clarity and archival purposes.

But I also have a question:  Do any of the "internal" OCR options result in the file being OCRed having those changes kept permanently?  I can't imagine that the single page "quick and dirty" OCR would do this as it would make no sense to even try because many documents are more than one page, and a partially OCRed file is as bad as one that still needs to be processed.  But I don't know much about the NVDA Advanced OCR (NAO) Add-On other than that it exists and what its primary purpose is.

If the file is to be used over and over and over again, and neither adds a text layer that sticks with the file, there are other options that do just that.  But they're not done by NVDA itself or an add-on.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


 

On Thu, Sep 8, 2022 at 04:06 PM, Howard Traxler wrote:
Do all the offered programs use the same OCR engine with just a little difference in the interface?
-
If we're talking just NVDA's built-in OCR and NAO, they might, but the GitHub page for NAO does not make it clear whether they're using the Windows OCR engine or another one.

There are a number of OCR engines out there, and I don't think anyone is actively reinventing that wheel anymore.  APIs exist for them.  But there are options other than Windows OCR that could be used.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


Rui Fontes
 

Hello!


NVDA's OCR and NAO both use Windows OCR.

TesseractOCR and NAPS2TesseractOCR, as name says, both use Tesseract OCR.


NVDA's OCR only performs OCR to the screen image.

NAO does that and also can OCR a image file, like PDF, JPG and  so on.

NAPS and Tesseract OCR do not make OCR to the screen image, but they can use a scanner to digitalize a paper document...


Rui Fontes



Às 21:06 de 08/09/2022, Howard Traxler escreveu:

Wondering about these OCR options:  Do all the offerred programs use the same OCR engine with just a little diferene in the interface?  Can't imagine that everyone wants to invent his own version of the wheel, as they say.


Howard

On 9/8/2022 10:33 AM, Brian Vogel wrote:
Topic retitled for clarity and archival purposes.

But I also have a question:  Do any of the "internal" OCR options result in the file being OCRed having those changes kept permanently?  I can't imagine that the single page "quick and dirty" OCR would do this as it would make no sense to even try because many documents are more than one page, and a partially OCRed file is as bad as one that still needs to be processed.  But I don't know much about the NVDA Advanced OCR (NAO) Add-On other than that it exists and what its primary purpose is.

If the file is to be used over and over and over again, and neither adds a text layer that sticks with the file, there are other options that do just that.  But they're not done by NVDA itself or an add-on.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


Howard Traxler
 

I've installed both the N A O and tesseractOCR addons on my win 11 machine.  They both are wonderful; and that tesseractOCR runs my scanner, too.  And it seems to run faster on the 11 computer; even the mechanics of the scanner.  Of course, thatt ain't possible; but it sure seems so.
Howard

On 9/8/2022 4:28 PM, Rui Fontes wrote:

Hello!


NVDA's OCR and NAO both use Windows OCR.

TesseractOCR and NAPS2TesseractOCR, as name says, both use Tesseract OCR.


NVDA's OCR only performs OCR to the screen image.

NAO does that and also can OCR a image file, like PDF, JPG and so on.

NAPS and Tesseract OCR do not make OCR to the screen image, but they can use a scanner to digitalize a paper document...


Rui Fontes



Às 21:06 de 08/09/2022, Howard Traxler escreveu:

Wondering about these OCR options:  Do all the offerred programs use the same OCR engine with just a little diferene in the interface?  Can't imagine that everyone wants to invent his own version of the wheel, as they say.


Howard

On 9/8/2022 10:33 AM, Brian Vogel wrote:
Topic retitled for clarity and archival purposes.

But I also have a question:  Do any of the "internal" OCR options result in the file being OCRed having those changes kept permanently?  I can't imagine that the single page "quick and dirty" OCR would do this as it would make no sense to even try because many documents are more than one page, and a partially OCRed file is as bad as one that still needs to be processed.  But I don't know much about the NVDA Advanced OCR (NAO) Add-On other than that it exists and what its primary purpose is.

If the file is to be used over and over and over again, and neither adds a text layer that sticks with the file, there are other options that do just that.  But they're not done by NVDA itself or an add-on.
--

Brian -Windows 10, 64-Bit, Version 21H2, Build 19044

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

~ Irving Babbitt


Rui Fontes
 

No, but you have two add-ons to do that:

tesseractOCR and NAO.


Best regards,

Rui Fontes
NVDA portuguese team


Às 13:50 de 08/09/2022, Richard Kuzma via groups.io escreveu:

Does anyone know if nvda has a function like the convenient ocr option in jaws where you can right click – do the applications key on a pdf and arrow down to convenient ocr with jaws and press enter and jaws will try and ocr the file and pull the text out of it?

Thanks,

Rich

 


Brian's Mail list account
 

The recent one I downloaded as an addon and apologies for my spelling uses tesseract ocr. It seems to work quite well, obviously, its dependent on the quality of the image its looking at. It certainly did not do a bad job on an old scanned manual for an audio limiter with two pages and I saved it but its not something I use every day, as I try to train the organisations to use the source file, but obviously legacy files like old manuals do need to be of good quality or the end result is in Klingon.


For the printed word, then I tend to use one of the many smart phone apps and just set that up as a rostrum camera and snap each page, or use a flatbed scanner and use one of the many engines on the image.
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: "Howard Traxler" <htraxler7@...>
To: <nvda@nvda.groups.io>
Sent: Thursday, September 08, 2022 9:06 PM
Subject: Re: [nvda] NVDA's built-in OCR Capabilities


Wondering about these OCR options: Do all the offerred programs use the
same OCR engine with just a little diferene in the interface? Can't
imagine that everyone wants to invent his own version of the wheel, as
they say.


Howard

On 9/8/2022 10:33 AM, Brian Vogel wrote:
Topic retitled for clarity and archival purposes.

But I also have a question: Do any of the "internal" OCR options
result in the file being OCRed having those changes kept permanently?
I can't imagine that the single page "quick and dirty" OCR would do
this as it would make no sense to even try because many documents are
more than one page, and a partially OCRed file is as bad as one that
still needs to be processed. But I don't know much about the NVDA
Advanced OCR (NAO) Add-On other than that it exists and what its
primary purpose is.

If the file is to be used over and over and over again, and neither
adds a text layer that sticks with the file, there are other options
that do just that. But they're not done by NVDA itself or an add-on.
--

Brian -Windows 10, 64-Bit, Version 21H2, Build 19044

It is well to open one's mind but only as a preliminary to closing it
. . . for the supreme act of judgment and selection.

~ Irving Babbitt





Aravind R
 

at present, i use tesseract addon or if that failes then, use NOW add
on. if both fails, then i go for mobile based facilities like envision
AI or Seeing AI or android based i-stem app as last resort. mostly
things get solved within these options.

On 09/09/2022, Brian's Mail list account via groups.io
<bglists@...> wrote:
The recent one I downloaded as an addon and apologies for my spelling uses
tesseract ocr. It seems to work quite well, obviously, its dependent on the

quality of the image its looking at. It certainly did not do a bad job on an

old scanned manual for an audio limiter with two pages and I saved it but
its not something I use every day, as I try to train the organisations to
use the source file, but obviously legacy files like old manuals do need to

be of good quality or the end result is in Klingon.


For the printed word, then I tend to use one of the many smart phone apps
and just set that up as a rostrum camera and snap each page, or use a
flatbed scanner and use one of the many engines on the image.
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: "Howard Traxler" <htraxler7@...>
To: <nvda@nvda.groups.io>
Sent: Thursday, September 08, 2022 9:06 PM
Subject: Re: [nvda] NVDA's built-in OCR Capabilities


Wondering about these OCR options: Do all the offerred programs use the
same OCR engine with just a little diferene in the interface? Can't
imagine that everyone wants to invent his own version of the wheel, as
they say.


Howard

On 9/8/2022 10:33 AM, Brian Vogel wrote:
Topic retitled for clarity and archival purposes.

But I also have a question: Do any of the "internal" OCR options
result in the file being OCRed having those changes kept permanently?
I can't imagine that the single page "quick and dirty" OCR would do
this as it would make no sense to even try because many documents are
more than one page, and a partially OCRed file is as bad as one that
still needs to be processed. But I don't know much about the NVDA
Advanced OCR (NAO) Add-On other than that it exists and what its
primary purpose is.

If the file is to be used over and over and over again, and neither
adds a text layer that sticks with the file, there are other options
that do just that. But they're not done by NVDA itself or an add-on.
--

Brian -Windows 10, 64-Bit, Version 21H2, Build 19044

It is well to open one's mind but only as a preliminary to closing it
. . . for the supreme act of judgment and selection.

~ Irving Babbitt











--


--
--
nothing is difficult unless you make it appear so.

r. aravind,

manager
Department of sales
bank of baroda specialised mortgage store, Chennai.
mobile no: +91 9940369593,
email id : aravind_069@..., aravind.andhrabank@....
aravind.rajendran@....


Steve Nutt
 

Hi Brian,

 

Not as far as I know with NVDA, but JAWS allows you to open the resulting file in Word, and it retains formatting as best it can. So it is definitely best out there for built in OCR to a screen reader.

 

All the best


Steve

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian Vogel
Sent: 08 September 2022 16:33
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA's built-in OCR Capabilities

 

Topic retitled for clarity and archival purposes.

But I also have a question:  Do any of the "internal" OCR options result in the file being OCRed having those changes kept permanently?  I can't imagine that the single page "quick and dirty" OCR would do this as it would make no sense to even try because many documents are more than one page, and a partially OCRed file is as bad as one that still needs to be processed.  But I don't know much about the NVDA Advanced OCR (NAO) Add-On other than that it exists and what its primary purpose is.

If the file is to be used over and over and over again, and neither adds a text layer that sticks with the file, there are other options that do just that.  But they're not done by NVDA itself or an add-on.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


 

On Fri, Sep 9, 2022 at 10:34 AM, Steve Nutt wrote:
Not as far as I know with NVDA,
-
Thanks.

Then for those who are dealing with documents, whether short or long (I've done documents in the hundreds of pages), that they know they'll be referring to again and again or want to share with others, I offer the following:  Using Tracker Software’s PDF-Xchange Viewer to OCR Process Image Scanned PDFs

It's not particularly labor intensive when you know that you need to reference a given PDF over and over again and only want to have to do OCR processing once.  Tracker also supplies multiple language packs beyond the "core four" at no charge.  I had a client once who was using PDF-Xchange Viewer to OCR process scanned PDFs in Swedish.

I've also encouraged my clients who were grad students, who kept being given image scanned PDFs created years ago, to OCR process them this way then give the resulting file back to their professor, asking them to replace the one they've been using that has no text layer, so future blind students don't have to do this at all.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


Howard Traxler
 

Brian, in your message quoted below, it seemed as though you were linking to an OCR app called PDF-Xchange Viewer.  I followed the link and downloaded a file called

A Short Tutorial on NVDA Screen Review
by Gene Asner.
A nice list of commands that I'll keep in my "help" folder but not an OCR program.  Was there one that you meant to link?

Thanks.

Howard

On 9/9/2022 10:47 AM, Brian Vogel wrote:
Thanks.

Then for those who are dealing with documents, whether short or long (I've done documents in the hundreds of pages), that they know they'll be referring to again and again or want to share with others, I offer the following:  Using Tracker Software’s PDF-Xchange Viewer to OCR Process Image Scanned PDFs

It's not particularly labor intensive when you know that you need to reference a given PDF over and over again and only want to have to do OCR processing once.  Tracker also supplies multiple language packs beyond the "core four" at no charge.  I had a client once who was using PDF-Xchange Viewer to OCR process scanned PDFs in Swedish.

I've also encouraged my clients who were grad students, who kept being given image scanned PDFs created years ago, to OCR process them this way then give the resulting file back to their professor, asking them to replace the one they've been using that has no text layer, so future blind students don't have to do this at all.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


 

Howard,

Yes, that was definitely a mistake.  I have no idea how the link associated with that hyperlinked material got changed (and I had to have done it by accident at some point).  Although I refer to the tutorial it downloaded on occasion, it's not this occasion.

The corrected hyperlink: Using Tracker Software’s PDF-Xchange Viewer to OCR Process Image Scanned PDFs

 

--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


Steve Nutt
 

Still doesn’t let you scan from a scanner or camera though, without any add-on. I think that’s where JAWS has the edge, though I would posit that OCR scanning is not really the function of a screen reader, beyond scanning the current screen, so I think that NVDA is right to not build this in natively.

 

All the best


Steve

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian Vogel
Sent: 09 September 2022 16:48
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA's built-in OCR Capabilities

 

On Fri, Sep 9, 2022 at 10:34 AM, Steve Nutt wrote:

Not as far as I know with NVDA,

-
Thanks.

Then for those who are dealing with documents, whether short or long (I've done documents in the hundreds of pages), that they know they'll be referring to again and again or want to share with others, I offer the following:  Using Tracker Software’s PDF-Xchange Viewer to OCR Process Image Scanned PDFs

It's not particularly labor intensive when you know that you need to reference a given PDF over and over again and only want to have to do OCR processing once.  Tracker also supplies multiple language packs beyond the "core four" at no charge.  I had a client once who was using PDF-Xchange Viewer to OCR process scanned PDFs in Swedish.

I've also encouraged my clients who were grad students, who kept being given image scanned PDFs created years ago, to OCR process them this way then give the resulting file back to their professor, asking them to replace the one they've been using that has no text layer, so future blind students don't have to do this at all.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt


Quentin Christensen
 

Steve,

That's a very contradictory point!

It's true that NVDA does not have the facility to let you scan a hardcopy document with a scanner directly, and we would also argue that is the realm of scanning software rather than your screen reader.  Indeed, part of your Jaws cost goes towards the OCR license, so you could use NVDA and purchase a program like Abbyy FineReader which would give you similar functionality.

We're always open to ideas on how to improve NVDA.  For something like this, we would either need a full OCR program which was free we could incorporate, or a paid add-on (or add-on which interfaced with a paid OCR program) which users could choose to use.

Quentin.

On Mon, Sep 12, 2022 at 11:20 PM Steve Nutt <steve@...> wrote:

Still doesn’t let you scan from a scanner or camera though, without any add-on. I think that’s where JAWS has the edge, though I would posit that OCR scanning is not really the function of a screen reader, beyond scanning the current screen, so I think that NVDA is right to not build this in natively.

 

All the best


Steve

 

From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Brian Vogel
Sent: 09 September 2022 16:48
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA's built-in OCR Capabilities

 

On Fri, Sep 9, 2022 at 10:34 AM, Steve Nutt wrote:

Not as far as I know with NVDA,

-
Thanks.

Then for those who are dealing with documents, whether short or long (I've done documents in the hundreds of pages), that they know they'll be referring to again and again or want to share with others, I offer the following:  Using Tracker Software’s PDF-Xchange Viewer to OCR Process Image Scanned PDFs

It's not particularly labor intensive when you know that you need to reference a given PDF over and over again and only want to have to do OCR processing once.  Tracker also supplies multiple language packs beyond the "core four" at no charge.  I had a client once who was using PDF-Xchange Viewer to OCR process scanned PDFs in Swedish.

I've also encouraged my clients who were grad students, who kept being given image scanned PDFs created years ago, to OCR process them this way then give the resulting file back to their professor, asking them to replace the one they've been using that has no text layer, so future blind students don't have to do this at all.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt



--
Quentin Christensen
Training and Support Manager


 

On Mon, Sep 12, 2022 at 09:20 AM, Steve Nutt wrote:
I would posit that OCR scanning is not really the function of a screen reader, beyond scanning the current screen, so I think that NVDA is right to not build this in natively.
-
And that's precisely what I think as well.

There seems to be this idea, among some, that a screen reader itself is supposed to be this Swiss Army Knife type tool that literally does anything and everything that any given user thinks it should be able to do.  I don't.  Screen readers exist to allow blind or visually-impaired users to access the same software that those of us who can see do use to do all sorts of things.

My multi-function is what I use to scan material (with OCR processing as part of the initial scan, and that's been common for years now) and the scanner software is what handles all of that.  If the scanning software is accessible, then the problem is already solved.  If it's not, then the issue lies with the scanning software, not the screen reader.

Having a "quick and dirty" ability to OCR what's on a screen is one thing.  Having the capability to act as substitute scanning software for all conditions is entirely another.
--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

It is well to open one's mind but only as a preliminary to closing it . . . for the supreme act of judgment and selection.

       ~ Irving Babbitt