OCR with Abbey very slow

Hi there,

I have noticed that the OCR process through Abbey Finereader seems to be unusually slow. I have nothing to compare it with, so I might be wrong, but it doesn’t feel right.

I have a very powerful machine (high end Mac Pro from last year) with Quad Core processors and 12 gigs of Ram. Still, Abbey takes anywhere between 10 seconds and one minute per page. It’s much slower than actually scanning the documents.

I noticed that the OCR process only makes use of a single processor core, so I figure either the Abbey component DT uses is very old, or Abbey just can’t deal with multi core processors.

Any thoughts about this?


The current license with ABBYY doesn’t allow optimizing OCR for multi-core processing, which would substantially increase the license cost. So my quad-core Mac has no advantage over a single-core Mac for those OCR processing flows.

Also note that DEVONtechnologies chose to save intermediate pages in the processing queue to disk, rather than to retain them in RAM. This tends to slow down OCR completion, but has the advantage that the computer is far less likely to be “choked” when a large document is processed. This makes OCR more of a “background” procedure, allowing the user to continue other tasks while OCR is going on.

Well the question is, is there anything we can do about it? I am scanning a lot of documents daily, and it takes all day to finish the OCR.

Is it possible to license the full Abbey Fine Reader and install it to replace the built in version?


Yes,you may install any OCR application – but without the internal hooks in DT Pro Office.

Are there any plans in the near future to add OCR multi-core processing as an extra? I have DTPO and would welcome the ability to use all cores for OCR to speed the process.

Seems doubtful, unless there’s a sufficient drop in the license cost::

I have asked in the suggestions thread that DEVONtech look into offering that as an upsell to people with DTPO. I would be interested in paying an additional license fee to activate multicore OCR.

1 Like