Output with Dragon
Thread poster: Nicholas Stedman
Could a professional technical translator with several years experience using Dragon give me an 100 % honest appraisal of his/her hourly output with voice recognition software for a text with no repetition. This would be for work of average difficulty from job reception to return to client of a finished and formatted document and therefore includes research time.
I am thinking of investing in such software but want to be sure that any change in my working practices is worth the effort.
[Edited at 2009-05-05 17:50 GMT]
| | Andrew Levine
Local time: 01:19
French to English
| My experience || May 5, 2009 |
Here is my experience with Dragon 9 Preferred. You are in luck because I actually timed myself for a while to see how much it was saving me. The short answer is that my hourly output for texts where it makes sense to use Dragon (including proofreading/review) has gone from about 900-1000 words per hour to about 1100-1200 words per hour. But this came after some false starts. There are some documents (see below) where it isn't as effective and so I don't use it. Counting both the time it has saved me and the time it has cost me, I would say it saved 150 hours of work in one year overall, so it has paid for itself about thirty times over.
A few key points:
1. It took some patience to calibrate the software for my voice. For the first five or six hours of talking time (spread out over a few weeks), I felt like I had gotten ripped off because it only got about 60%-70% of the time. However, as the manual instructed, I corrected my errors using the voice-select and correct functions whenever possible.
2. As my files were saved, the accuracy steadily increased. After about a month of use (about 10 hours talking time, during which my output was about 80%-90% of what it would have been had I not been insistent on getting Dragon to work) it had reached a very high-quality plateau, which it has remained at for over a year now without further improvement.
3. That plateau is about 95% accuracy, with the words appearing on the screen about 20% to 30% faster than I could type them. Also, any words which the software recognizes correctly (i.e. not homophones or soundalikes) are spelled with 100% accuracy, which can save some time when using a spellchecker. You have to check to fix those ~5% of the words you get wrong, but this only causes me to spend about 10% more time proofreading than normal.
4. Long sentences are now recognized almost flawlessly (~98%) and are entered much faster than I could type them. A huge number of sentence fragments, and text peppered with rare acronyms (e.g. it is good with everyday acronyms like GPS or NAFTA or MSNBC, but teaching it something like "OEA" or "GPRQ" takes time) are entered no faster than typing, and in some cases you may lose time correcting them.
I am overall very satisfied with Dragon (see the part above where it has paid for itself thirtyfold!), and if you have a tendancy to translate technical documents and the willingness to invest time at the beginning for a return later on it is an excellent choice. If you do not use a CAT tool, that is even better because the time savings are much greater that way than when going from (say) Wordfast to Wordfast+Dragon.
Oh, and get a good headset, it's worth it.
[Edited at 2009-05-05 23:28 GMT]
| Thanks for your very helpful answer || May 6, 2009 |
Many thanks Andrew for your extremely helpful answer. I can now make a direct comparison with my own output.
| | Alistair Gainey
Local time: 06:19
Russian to English
| I agree with Andrew || May 9, 2009 |
I'd also like to add that there are considerable physical advantages to using Dragon as well. Typing several thousand words a day leaves my fingers knackered, but I don't have that problem at all with voice recognition software. You do have to train the thing properly though (it's worth settting aside a few hours for this), and a good headset is indeed vital. Eventually, you might even find 1200 words an hour to be an underestimate (even including proofreading) for your output for texts on subjects familiar to you. Let's say you want to do 1500 words an hour. If the time you devote to proofreading is 1/3 of the time you devote to translating, that means you need to do 1500 words in 45 minutes. That makes about 333 words every 10 minutes. I did a quick experiment just now to see how many words I could produce just by talking for 10 minutes, and got over 350 -and that was while I was trying to think of things to say. True, 1500 words an hour may be an optimistic target if you've got a lot of research to do as well, but 1200 words is by no means unrealistic.
To report site rules violations or get help, contact a site moderator:
You can also contact site staff by submitting a support request »
Output with Dragon
|SDL Trados Studio 2017 Freelance|
|The leading translation software used by over 250,000 translators.|
SDL Trados Studio 2017 helps translators increase translation productivity whilst ensuring quality. Combining translation memory, terminology management and machine translation in one simple and easy-to-use environment.
More info »
|Déjà Vu X3 |
|Try it, Love it|
Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market.
See the brand new features in action:
*Completely redesigned user interface
*Inline spell checking
More info »