Browse access-uk: This Month's Archive • Main Archive Page • Related posts • Previous by Date • Next by Date
[access-uk] Re: TALKING PICTURES

From: "Tristram Llewellyn" <tris-l@xxxxxxxxxx>
To: <access-uk@xxxxxxxxxxxxx>
Date: Thu, 3 Aug 2006 11:00:29 +0100
Hi Ian



This is a question I would love to fully answer but cannot.  The 
question/argument you riase comes up and is quite a good example to take up 
because it seems to fit some commonsense notion that blending existing 
technologies and re-packaging them at inflated price is what has been done 
here.  Commonsense seems to dictate that this is what is going on with the KNFB 
reader, and it sounds even worse if an American blind charity is involved but 
not dishing these out to their users at reduced or no cost.  This is a question 
based on a little knowledge of technology; therefore, the answer will also have 
to be of a technical nature, even as it is it will not entirely meet the 
question..  

 

I will try to elucidate on this specific example at a technical or engineering 
level.  The reason is that to take in a whole industry, economic structures, 
economies of scale, financial obstacles to development, short term behavior of 
western venture capital, specialist markets would be beyond my qualification 
therefore I cannot answer the whole equation.  The point about the Gates trust 
is interesting, I confess that I do not know why not, but it probably depends 
upon its focus they have.  There is an issue about inward investment and the 
economics but I fear the answer is probably not as cut and dried as anyone 
would like to think, of course such companies know the cheaper something is, 
the more people are able to buy into it, but just knowing that doesn't actually 
make it happen.  Just as a brief diversion if I may, I can personally remember 
having some tiny involvement in it, when organizations for the blind were 
approaching people to produce standards and lobby manufacturers and developers 
to produce an accessible DAB radio.  When after approximately five to six years 
had elapsed a model has come out, it was met with what Pure called 
"disappointing sales" and that was your cheap everybody get in there price 
sitting right at the other end of the price spectrum.  I have a notion from 
seeing from the inside of some of these difficult issues but I do not think I 
am qualified (at least yet) to give any public analysis of it.  I can point you 
in the direction of what is going on with the reader itself at some technical 
level that would help to explain what is behind it rather than what is apparent 
on the surface.

 

The answer in short to the initial question is that you would be quite right to 
assert that the price tag of KNFB reader is not equal the hardware alone, but 
it is not all hardware.  Commonsense might dictate that slicing a camera, PDA 
and some OCR software together into a box you have something like a KNFB 
reader.  This is an attractive proposition isn't it, but anything that seems to 
good to be true probably is just that otherwise we'd all be at our kitchen 
tables making KNFB readers.  Unfortunately, that is a cooking methodology, 
mixing ingredients; it is not equal to the task of making a portable text 
acquisition device because it doesn't aim to solve the problems inherent in 
such a device and that is where you need to start. 

 

The problems thrown up by creation of a free moving portable text acquisition 
device are quite different from those of a flatbed scanner and this is why you 
cannot just get an OCR package to run a KNFB reader.  Consider anecdotally that 
even some sighted people can take awful pictures with a camera that technically 
is quite capable of taking much better pictures.  Even with adequate to full 
sight, not everyone knows how even with the features at their disposal take a 
picture that technically compensates for issues of lighting, focus, depth of 
field etc.  Then, if you then chuck in a blind user, things might get really 
unpredictable.  The subject of their picture is relatively fine print, probably 
not in ideal lighting conditions, the camera can shake whilst taking the 
picture, they may be at an awkward angle, the spine of a book causes curvature 
of the text in places and we have a scanning problem ready to chew on which is 
quite different from a flatbed scanner.  With a flatbed scanner, you have a 
relatively controlled situation with known distance to the medium you are 
scanner, controllable lighting conditions.  A portable reader is somewhat 
different.  If you don't take my word for it take a look at this interview with 
Ray Kurzweil himself.  The target audience is mainly the engineering community 
but may still prove informative and fascinating in parts. 

 

http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=12979

 

If you read the article, you will begin to understand where some of that extra 
cost beyond the hardware has gone.  If you haven't yet and you are still in the 
dark, it is the software that is driving the thing that is costing a 
significant amount of money.  Even with NFB behind with some funds, 
encouragement and other support.  Specific technical issues create significant 
challenges that have to be met to create a workable product.  Some entirely new 
bits of software have to be created which are not immediately obvious to the 
user that is critical for the reader's operation.  The software that describes 
what the user is taking a picture of was a significant and new task.  Why do we 
need it, because the user taking the picture has no or little sight?  How do 
you tell a person who has limited or no direct sensory experience of picture 
taking what is there to correct any problem.  This would be amazingly difficult 
and has cost many person-hours of time to work out as well as it is has been 
done. 

 

Ray Kurzweil has an amazing track record, and has had a long association with 
NFB in the states.  Though it is difficult to play the game of historical "what 
if", it may be that if it were not for his omni font character recognition 
techniques we might not be where we are now.  He is optimistic about future 
prospects for what this technology can do in the future.  His first scanner for 
the blind cost $25,000 in 1976 as a starting point, we are starting with 
something somewhat cheaper now, cheaper than the famous Opticon when it was 
first introduced costing something like the £3000 in it's day but closer to 
£10000 in today's money.  This is all of course relative and the ratio of price 
drop is going to be different, but it is a start.


 

Regards.

Tristram Llewellyn
Sight and Sound Technology
Technical Support
www.sightandsound.co.uk
Follow-Ups:

[access-uk] Re: TALKING PICTURES
From: Marie Baisez
References:
- [access-uk] Re: TALKING PICTURES
  - From: Ian Macrae
[access-uk] Re: TALKING PICTURES

Other related posts:
