Connect with us

Mobile

Dragon Medical 10-Week Three

Published

on

Dragon Medical 10 has surpassed my expectations! I didn’t expect to be writing this first sentence since I am a long-time user of Dragon, going back almost 1 decade. I have historically found that each new version of Dragon has some improvements but not necessarily great improvements.

I have developed certain expectations regarding Dragon’s accuracy as the program has changed over the years. I have learned about microphones, tweaking my computer for optimal performance and other things in my quest to have perfection.

I never thought I would see the program developed to such a degree that it really is almost perfect with speech recognition. Now, I seem to be having 99% accuracy in daily usage. Yes, I have a great microphone. I also have a Motion 1600 tablet computer that really isn’t state of the art.

The words do not go down on my tablet screen as quickly as they do on Dragon’s promotional literature. I guess if I did have the optimal computer that that also would occur. This has not necessarily presented to be a problem for me. I have found that I need time to think about my thoughts and the lapse of time after I speak until the words are transcribed is good for me.

I am using a Sennheiser headset. This is a wired headset and it would be great to use a Bluetooth headset but I usually use a Bluetooth keyboard. I can honestly say that I have yet to see a Sennheiser microphone that does not work exceptionally well with speech recognition.

Medical terminology already embedded in the program has really expedited my increase in accuracy. I am not training any of the medical terms as I had in the past with Dragon professional. Although many medical words are in the vocabulary of the professional version it often times did not recognize them without training.

So what are my limitations with this program? Well, I am disappointed about the fact that when I say “select a certain phrase” often times is transcribed as “slight a certain phrase.” This may be a reflection of not having a fantastic tablet that brings out the best in the program. I think this lack of focus may be more a result of a small antiquated chipset. I have not noted others experiencing this issue.

This week should bring some other comments regarding speech recognition from both Truc and Craig. It will be very interesting to see what they have to say. Stay tuned!

7 Comments

7 Comments

  1. Joe

    10/19/2008 at 8:45 pm

    I don’t know about with Dragon, but I attempted a bluetooth headset with Vista’s built in speech recognition, and the bluetooth mic just doesn’t have the sound quality for it to recognize anything at all, so you’re better with the wired.

  2. cphickie

    10/19/2008 at 10:13 pm

    There’s an interesting discussion over on emrupdate.com about Dragon 10 Medical. Apparently (if I read this correctly) you cannot use the much cheaper Dragon 10 Preferred version in most EMRs now (unlike with prior versions of Dragon) due some sort of software restriction put in place by Nuance. Many of the doctors in this discussion are not very happy about now having to shell >$1000 for the DNS 10 Medical version versus the ~$200 for the preferred. The link is: https://www.emrupdate.com/forums/t/16936.aspx

  3. JohnL

    10/19/2008 at 10:38 pm

    John,

    Thank you for your update on this.

    Which specific Sennheiser headset do you have. I need to get one and would rather get one that has been proven.

    Best regards

  4. RMathew

    10/19/2008 at 11:19 pm

    I also had the issue with selecting text that you describe. It’s been awhile since I used Dragon (my workplace isn’t conductive to talking aloud about our material) but I recall it was my biggest annoyance with the program. The software is great and has surprisingly good accuracy, but afterwards I always had to take out the keyboard and correct everything by hand. I think they need another word than “select” – something that is not too hard to separate and ID. Perhaps “mark” instead.

    I also had issues with bluetooth headsets – the audio was just not clear enough for Dragon – even with good sound from the headset the slight distortions created by the connection itself seems to mess it up. If anyone finds a good bluetooth headset for Dragon I’d be quite interested in trying it again.

  5. Phillip

    10/20/2008 at 7:12 am

    apparently there is a workaround to the EMR problem with Dragon naturally speaking pro version. If you change the name of the file from yourEMR.exe to something else youremrrenamed.exe it may work.

  6. Michael

    10/20/2008 at 11:19 am

    I often work with physicians to help improve speech recognition accuracy and I wanted to take a moment and expand on your statement (â€Ŕselect a certain phrase” often times is transcribed as â€Ŕslight a certain phrase.”) The general rule for a misrecognized word is, â€Ŕif a word is not recognized accurately often, then train the word.” Your article points out that sometimes this doesn’t work, so here’s a few tips on how to handle words that are trained but still get misrecognized.

    We’ll use the phrase, â€ŔThe patient.” 1. If the phrase â€Ŕthe patient” results into â€Ŕa patient” then don’t train the word â€Ŕthe” train the phrase â€Ŕthe patient” as one word. You’ll find that in any given document the word â€Ŕthe” is recognized correctly in many places but for some reason not with a certain phrase, in this case â€Ŕthe patient.” Therefore, training that phrase may fix the issue. 2. If the phrase â€Ŕthe patient” results into â€Ŕthe Asian” even after you train the word patient, consider training the word â€ŔAsian” and you may find that the word â€Ŕpatient” doesn’t result into â€ŔAsian” any longer. 3. Alternatively, let’s say the word â€Ŕthe” gets recognized as â€Ŕa.” Ask yourself if this happens often at the beginning of a sentence? If yes, then the microphone may not be in record mode before you begin speaking. If you break down the word â€Ŕthe” you’ll find that â€Ŕthe” is a sequence of the sounds â€Ŕth” followed by â€Ŕa.” The same holds true for this phrase, â€ŔPatient was seen by…” resulting into â€ŔAsian was seen by…” The â€ŔP” in patient didn’t get recorded. This is referred to as clipping and is hard to see since it’s only missing by less than a quarter second. Consider this sequence of steps to help prevent clipping: Press the record button, inhale, then begin to speak.

    For the issue you have with â€Ŕselect” resulting into â€Ŕslight”, I would recommend that you first train the word â€ŔSlight” and if that doesn’t work, then train the phrase â€Ŕselect a certain” replacing â€Ŕa certain” with actual words from the phrase you are trying to access.

  7. Wayne

    10/20/2008 at 3:19 pm

    It’s really a problem that Nuance dictate what you and can’t do with a program. It didn’t know about it till I tried it.

Leave a Reply

Your email address will not be published.

As an Amazon Associate I earn from qualifying purchases.