Faxing in Synapse - which product to integrate ?

Discussion in 'Feature: Requests and Planning' started by Graham, Jan 19, 2007.

  1. Graham

    Graham Developer Staff Member

    I've thought about this a little, and my idea is that I will write a couple of helper agents. These will be launched by the synapse server and will terminate when the synapse server also terminates.

    There will be a helper agent that monitors a "fax" directory. Any tiff, g3, or pdf files found there will be inserted into the fax queue.

    Another agent will be used to upload the cache listener files to S3, and the split remr gbak files.

    And a final agent will scan all the files looking for those that have not been ocrd. It will ocr them and insert the ocr text.
  2. Graham

    Graham Developer Staff Member

    Note that the "fax" and "ocr" agents already exist as stand-a-lone applications .. and so I just need to rewrite them to be gui-less, and just tray icon apps.
  3. Graham

    Graham Developer Staff Member

    Another candidate as an agent is the Medisoft bridge ... which runs as a GUI at present.
  4. Jerry

    Jerry Administrator Staff Member

    Would the ocr agent still require the online scanning service? I don't currently use Autofile, but I'm considering it as I'm starting to use fax receiving capabilities more.
  5. Graham

    Graham Developer Staff Member

    Yes. The ocr agent would wrap an existing 3rd party commercial web service.

    But it's optional.
  6. Graham

    Graham Developer Staff Member

    The FaxQ agent is now available for testing ...
    1. Upgrade your client to 80
    2. In Settings/User/Fax Inbox - fill the first two fields in
    3. Save the FaxQ settings
    4. Upgrade your server
    5. Place some files in the fax watch directory
    6. Check your Inbox/Faxes tab and see if they appear

    Looks like my original app OCR'd the faxes as well.... using Tesseract, so that's an enhancement I need to add back in.

    Also the [last scanned occurred at] button is not done yet.

    The FaxQ agent is started when Synapse-Server starts ... so you can't invoke except by restarting the server at present.

    It should terminate within the polling period after the server closes down. You can close it down manually by using the menu from the system tray icon.
  7. laumansm

    laumansm New Member

    Tesseract only yields modest results. Is there any way one can implement a third party OCR application such as Acrobat (or whatever comes with your scanner) - and not use the online OCR service nor Tesseract? Would there be any benefit to pre-OCR my documents when scanning them in?
  8. Graham

    Graham Developer Staff Member

    Tesseract is able to give you the flavour of the document.

    Acrobat does not have OCR facilities. It most likely uses a 3rd party tool.

    We could look at using a 3rd party tool such as Leadtools. The cost of their software is $1995. See http://www.leadtools.com/SDK/Document/Document-Addon-OCR.htm#buy , and there are then additional licensing fees on top of that. If you would like to donate an SDK, I can look at it ...

    The OCR web service I use will cost approx 2c a document.
  9. Graham

    Graham Developer Staff Member

    Omnipage Capture SDK 15 ( the version which has just been superceded by 16 ) is $4895 with additional run time license fees.

    Again .. I can look at it if someone donates a copy to me.

    See http://www.nuance.com/omnipage/capturesdk/
  10. Graham

    Graham Developer Staff Member

    Leadtools also has a web service for OCR. See http://www.leadtools.net/ecom/hostedplans.aspx

    charge is $55 setup and $19.95 per month.

    So, if you were scanning more than 1000 pages per month, then leadtools webservice would be cheaper than the service I am using.

  11. Graham

    Graham Developer Staff Member

    Since I have not received any reports on how this is working .. I am going to assume that it is working fine.

    I will now work on the S3 backup Agent.

  12. Graham

    Graham Developer Staff Member

    Nuance just returned my enquiry ... they are charging $1595 per license per server if I were to use their SDK with Synapse-Server.
  13. laumansm

    laumansm New Member

    Ja, Heidewitzka (as the Germans say)!

    I did not know that this OCR business is so involved. I was wondering whether there would be a way to import either a .doc document or .txt or searchable .pdf etc. and bypass any online OCR service. My scanner allows me to scan and OCR at once and generate any of the above text-containing files. I believe most newer scanners come with that sort of funcionality. It would be great to leverage that functionality this way.

  14. Graham

    Graham Developer Staff Member

    Your faxes come with included OCR'd text?

  15. laumansm

    laumansm New Member

    No. Oops, I guess that this is a fax related thread. All my scans can be OCR'ed automatically at the time of scanning. My faxes come as pdfs (or tif, if I want to) and automatization of OCR would be a bit more tricky but is doable also.
  16. Graham

    Graham Developer Staff Member

    Well, if you've tried out the FaxQ software, you can see that automated OCR is occuring already .. it's just not commercial quality.

    If you want best quality .. you need to purchase commercial SDK and licenses, or be prepared to pay for an OCR web service.
  17. Jerry

    Jerry Administrator Staff Member

    Graham, have you incorporated any auto deskewing/despeckling or image improvement/enhancement into the Fax Server agent? I'm assuming not, but just checking. That sort of thing can help OCR accuracy quite a bit. The nuance price is intriguing. It wouldn't take too many folks to want more accurate OCR to pay for it ... but I guess it is not unlimited or royalty free at that price, just running on one server?
  18. Graham

    Graham Developer Staff Member


    No, I haven't. Usuallly faxes in my experience don't need deskewing.

    The Nuance price is per Synapse-server.

  19. Jason

    Jason Developer / Handyman Staff Member

    I'd pay top dollar for OCR for AutoFiling.

  20. Graham

    Graham Developer Staff Member

    I think you'd be the only one.

Share This Page