Converting your printed book to Kindle through optical scanning? Think again.
Last year when I was first considering putting my back-list titles on Kindle, my UK agent, the illustrious Robert Kirby (who also represents Ricky Gervaise, Anthony Horowitz and Dawn French) mentioned that a number of agents had been pitched by Amazon at the London Book Fair about a special book-to-Kindle service.
Part of this involved optically scanning books (known as OCR in the trade) and then converting them to ebook Kindle format. However, at the time I'd gained good input from other authors and also I had all of my books on electronic files on my computer. In other words, the kdp conversion process was easy.
So, I went at it without using this specialized Amazon service - until I got to the fourth book in my list - The Shadow Chaser. For some strange reason, I had two thirds of the book on file but the rest was missing. I phoned Penguin to see if they had copies on file - no luck.
So in the end I was stuck with using OCR for the final third of The Shadow Chaser. The converted copy that came back was a nightmare. At least five misread words per page, sometimes far more, with on occasion entire lines mangled and rearranged. It took me ten days solid to go through it and take out the gremlins. A whole book I envisaged would take nearer a month. I got to thinking that it might have been quicker and easier re-typing the whole thing.
The problem with OCR is that it takes the nearest approximation to the word it 'thinks' it sees. Check out the same problems found with an early Kindle edition of Game of Thrones. The region known as Dorne was misread as Dome. The word 'don't' with a thick apostrophe might be misread as donut, etc. This appears to be a common problem with the OCR system.
Having worked so tirelessly to get rid of these errors, it appears that two or three still remained and only recently I found myself going back into the script to correct them. If any other authors have experienced the same problem with OCR, I'd be pleased to hear about it. Or indeed readers finding Kindle editions with multiple errors from mainstream publishers where you'd expect better.
Last year when I was first considering putting my back-list titles on Kindle, my UK agent, the illustrious Robert Kirby (who also represents Ricky Gervaise, Anthony Horowitz and Dawn French) mentioned that a number of agents had been pitched by Amazon at the London Book Fair about a special book-to-Kindle service.
Part of this involved optically scanning books (known as OCR in the trade) and then converting them to ebook Kindle format. However, at the time I'd gained good input from other authors and also I had all of my books on electronic files on my computer. In other words, the kdp conversion process was easy.
So, I went at it without using this specialized Amazon service - until I got to the fourth book in my list - The Shadow Chaser. For some strange reason, I had two thirds of the book on file but the rest was missing. I phoned Penguin to see if they had copies on file - no luck.
So in the end I was stuck with using OCR for the final third of The Shadow Chaser. The converted copy that came back was a nightmare. At least five misread words per page, sometimes far more, with on occasion entire lines mangled and rearranged. It took me ten days solid to go through it and take out the gremlins. A whole book I envisaged would take nearer a month. I got to thinking that it might have been quicker and easier re-typing the whole thing.
The problem with OCR is that it takes the nearest approximation to the word it 'thinks' it sees. Check out the same problems found with an early Kindle edition of Game of Thrones. The region known as Dorne was misread as Dome. The word 'don't' with a thick apostrophe might be misread as donut, etc. This appears to be a common problem with the OCR system.
Having worked so tirelessly to get rid of these errors, it appears that two or three still remained and only recently I found myself going back into the script to correct them. If any other authors have experienced the same problem with OCR, I'd be pleased to hear about it. Or indeed readers finding Kindle editions with multiple errors from mainstream publishers where you'd expect better.
RSS Feed
