Updated: February 4, 2016
What's the defining characteristic of greatness? The ability to accept feedback. In other words, when you ask a reviewer to test your product, and they come back with a less than positive article, you do not balk. Instead, you read through, figure out whether the criticism was based on hard facts, and improve.
Which is why I was more than happy to give Investitech a second chance when they contacted me again. My initial impression with PDF Converter 9 was rather lukewarm, focused on the product's high price, slow conversion speed, various errors when converting to DOCX and ODT, and a few other niggles. This time, we have version 10 at our disposal. Let's see how it fared.
Install, launch, convert
The user interface remains largely unchanged. You load documents, with a thumbnail preview shown in a sidebar to the left, and conversion options in a horizontal toolbar above the main area. However, they will be grayed out until you select one or more objects in the loaded files.
As before, the programs comes with a trial. The price for a full license stands at the unchanged USD99.95, but there's a new option, a USD34.95 30-day subscription, allowing you to use the product for a month. This can make sense for people with a limited budget and limited, one-time needs for file conversions. In other words, if you do not expect to be using software of this kind that often, the full price might be too steep. Then, you could grab the short license, work furiously for a few days, and then let it rest.
I decided to start with the same document I used before, my Linux kernel crash book, because it gives us an excellent baseline and comparison to version 9. Previously, the program did struggle a little bit with this file. Now, Able2Extract 10 offers a different experience.
Much like the last time, if you try to convert the file to a format that does not have an interpreter installed, e.g. Microsoft Office, you will get a warning, but the program will still continue with the conversion. Eventually, the process will fail, and you won't have the Word document available.
I find this weird. Either the conversion should not start at all - but that means limiting users, or it should run and not throw any exceptions, leaving it to the user to find a computer with the necessary software to open the relevant files. However, as far as the conversion goes, it was significantly faster this time. We are talking seconds rather than minutes. I checked the CPU utilization while the conversion was underway, and the four cores were all merrily buzzing. So I guess there's been some neat optimization under the hood. Timewise, the results are now perfectly acceptable, even though the clunkiness of the missing Office install conversion still remains.
I tried the OO conversion next. It still reads OpenOffice rather than LibreOffice, and the process was slower than Microsoft Word. However, the book was converted, and with a very high fidelity, mind. All the images have decent resolution, the page count is correct, text wrapping and line length are correct, and the converted file has all the charm of a more than acceptable result.
Converting this one file wasn't good enough to satisfy my needs. So I decided to test another document. This time, I went for A Mathematical Theory of Communication, the awesome 1948 Bell Labs paper by Shannon, which set the foundation for all modern communication, including the Internet.
This is a tricky document, as it comes with squiggly typewriter fonts, lots of noise, mathematical equations, and such, making the OCR challenge all the more appealing. Indeed, PDF Converter 10 stalled for a good minute loading the document, and I almost thought that we were going to have a hiccup, but then it completed just fine.
The conversion to the ODT format was also quite decent. Again, it took a long while, and I assume part of the blame goes into the OCR process, of trying to figure out all the little bits and pieces. The end result was a file with double spacing between words and mathematical equations that were all rather scrambled. Not bad, though.
If we compare to some other products we had in the past, namely Tesseract for Linux and friends, YAGF, and finally FineReader for Windows, we can see that the domain of OCR is never quite a plug-n-play exercise, and that you need a lot of time and patience to train programs. Then, there's also the problem of noise, alignment, font, angle, and whatnot. In this regard, I don't think there's anything cardinally bad to be said about PDF Converter. But it remains a forever challenge to nail OCR down perfectly.
You can possibly get better results if you play with the OCR settings and output options, but I have not tested those, so I can't really claim results any which way. Then, there are a few visual glitches here and there. For instance, while converting files, the progress bar popup is not centered, making my OCD pixels shiver in terror.
Able2Extract PDF Converter 10 is a better product than its predecessor. The interface remains the same, and it's never been an issue. Simple enough to use and navigate, if a little dull. On the bright side, the conversion speed and quality have massively improved, and that's one of the key points, plus there's a more affordable pricing option for less keen users of this type of software.
The Microsoft Office missing install problem still remains, and it should be sorted out, there are a small cosmetic issues in the GUI, and the challenge of making the software as accurate as possible when working with less than perfect documents remains. However, all considered, the program that was only marginally interesting is now something you might want to check. The trial is good enough to get the sense of quality, speed, efficiency, and overall practicality, and the 30-day subscription should satisfy the needs of most people other than professionals. 7.75/10. You might want to take a look.