Tuesday, August 30, 2011

Speech Recognition With General and Medical Vocabularies Revisited


A while ago, back on May 10, 2009, I posted an article that discussed speech recognition software in general and Dragon Medical 10 in particular. Click here to navigate to that post.
In the two years plus since then, the field of speech-to-text translation (and the hardware available for speech-to-text translation software to execute on) has advance. This can be seen in the research reports published at the Special Interest Sites that you can navigate to from the links at the top of the right-hand column of this blog.



Reflecting these advances, Vendors, Nuance Communications Inc., the maker of
Dragon NaturallySpeaking and Dragon Medical Practice Edition, in particular, have been able to release improved versions of their products: The next version of Dragon Medical will be called “Dragon Medical Practice Edition” and will feature the Dragon NaturallySpeaking 11.5 Speech Engine. I have found it to be faster and more accurate than any previous version of Dragon. (The most recent versions utilize new hardware features such as multithreading and multicore CPUs, and they double the sampling rate and more).

That said, it's still important that you use any speech-to-text translation product with a good sound card and a good microphone.


Click here for a downloadable Product Sheet

Click here for a Feature and Specs Sheet

As is my wont, I recently installed a pre-release copy of
Dragon Medical Practice Edition hurriedly: that is, I wasn’t very careful when I built a profile (trained Dragon) and then gave Dragon an initial test run by speaking into a 10-year-old, $5.95 microphone that happened to be connected to my sound card at the time. The idea was simply to familiarize myself with this new edition a little before using it seriously later. To my astonishment, even with this primitive set up, Dragon Medical Practice Edition converted my speech to text without a single error throughout a 5-minute run (using the General vocabulary).

There's an awful lot of information about this new version to sift through. Nuance's site (click
here) is a good place to start.



{ Click on the image above for a larger view }



As a footnote, I want to add that systems that convert spoken words to text are not just of value to professional practices. I have an acquaintance who has used this software for over a year as he has had a stroke and finds it difficult to type. He speaks well of it.