Speech dictation

Thanks guys. I agree that it “just works”, but we should document how to set up Dragon (non-free) and Simon (free) with our system. We are not advising the users to use it, it’s just there to be a helpful guide. This is similar to how we have the Twilio (non-free) documentation in place on the wiki

Is this agreeable?

Seems like we’d just be rewriting their documentation. Not worth it, in my opinion.

Hi Robert,

We don’t want to rewrite their documentation, but we should have a wiki page that points to the documentation and gives advice on setting this all up in the context of OpenEMR.

As far as Simon goes (the free/open source option). Oh my gosh is the documentation bad and the solution is still not working for me! I really hope to get it working so low resources user can make use of it (Dragon may be too pricey). Perhaps the documentation I create will be helpful to the Simon team.

Anyways, I am volunteering to do this project once the cTAKES project is out of the way. I agree with the comments that Dragon is pretty much plug and place, but I do want to make sure that a) it is known that Dragon and Simon are the options and b) links and helpful tips are provided so the user is up and running.


Is this agreeable?

If you’re up for doing the work, it’s fine with me

1 Like

Documentation is always a good thing :slight_smile:

1 Like

Cool, cool.

I am going to get the cTAKES and LightSail project done first, but I’ll totally sign on to do this!

1 Like

One thing to add to this discussion is that out of the box, Dragon has basic functionality for dictating to fields on web pages which is somewhat limited. There are add-ons and extensions for the various browsers that allow more Dragon functionality, such as the ability to correct errant dictation.
Apparently, the extension for Firefox recently became obsolete (see http://www.nuance.com/products/help/dragon/dragon-for-pc/enx/professionalgroup/main/Content/Web/about_firefox.htm), though I don’t recall it making a big difference when I had originally added it. Chrome and Internet Explorer both have functioning extensions for Dragon, but I opted not to use them because I know that OpenEMR is optimized for use with Firefox.
If one is doing very basic dictation, such as SOAP notes with little to no formatting, Dragon will be fine for dictating directly onto the browser fields. But, for any more sophisticated formatting, such as multiple font sizes and styles, fixed and hanging indents, etc., which make a dictated note look professional and easy to review, it is necessary to dictate one’s notes into a Word document, then copy and paste it into OpenEMR from there. I created some Layout Based Forms which are just a single field with no size limits, and I paste the Word document contents into these. There is some tweaking I have to do after this, such as changing the Format settings in LBF from “Normal” to “Normal (DIV)” in order to maintain the formatting exactly as it was in Word.
All of this means numerous extra steps and mouse clicks, but I have created several AutoHotKey macros that perform all of those extra steps with the push of one button. This, to me, provides the best of both worlds: the ability to have fully functional dictation by using Word, and the macro which saves most of the time it takes to move the text from Word to OpenEMR.
I should add that I’m using Dragon Medical v.10, which is several years old, but has amazing accuracy. It is possible that the newer (and waaaay more costlier) versions of Dragon Medical have improved browser functionality, but they still require installation of the browser extensions. I doubt they’ll ever get the browsers to do half of what Word can do with Dragon.

Hi ptalerico,

Thanks for this rich information.

I am glad to hear that you have found Dragon to work well with the system.

I am actually interested in solutions that don’t require Word -> Browser, but I realize that Firefox may not be best for this. Fortunately, OpenEMR works flawlessly in Chrome so there may be more options there.

As far as dealing with the user accidentally “backing out” of the web page (and losing all work), I wonder if temporarily caching the text in session/local storage as it comes in would be appropriate. Not sure if there a HIPAA issue with this, however. Also have to consider the scenarios in which it is best to delete this cache.


I’m glad to hear OEMR works fine in Chrome. I will try doing some Dragon dictation directly into the Chrome browser and can post observations, limitations, etc.
I can definitely attest to the frustration of accidentally backing out of the web page. I lack the technical expertise to comment on temporarily caching the text, but would anticipate that, as long as everything is stored locally and the cache can regularly be cleared, there shouldn’t be any HIPPA issues. (However, I certainly would defer to anyone on this forum who has more than my rudimentary training in HIPPA compliance).

1 Like

Has integration to Nuance been instantiated into OpenEHR?

It will work, but it’s not open source so we can’t officially support it.

If you are interested, our group can use sponsorship to integrate a free and open source solution called Simon.


Matthew Vita
OpenEMR Administrator

Hi Matthew,

Some background: I am on a 3-year contract with the Western Cape Department of Health. They have a combination of “burning platform” and Health Informatics innovation drivers pushing them to dramatically increase the impact envelope of ICT in healthcare. I came onboard after I briefed them on biometric Individual Identity Management. I am aligned with RightPatient/M2SYS. My initial work object was to assess the potential to deduplicate their heavily duplicated PMI - can’t be achieved due to the significance of the duplications. I am proposing a “ground zero” PMI rebuild using a biometric/Individual Identification as a Service platform (b/I2aaS) which will populate a transitionary Summary Care Record while the elephant in the room - 3 disparate PAS/EPR applications are currently deployed - is addressed. I am recommending OpenEMR as the single source solution. I am busy building the business, functional and technical case for taking this approach. The challenge I had was the at the point of service digitisation of consult notes. Keyboarding is not an option given the time constraints placed on Clinicians in terms of Patient throughput. I was reviewing the OpenEMR blogs and discovered your speech to text work. I attempted to use Dragon and IBM’s product more than a decade ago but gave up as they were “Klunky” and erratic.

Funding: I can analyse and present an investment case for the Department to indirectly support your efforts. The route of least resistance is that of an “add-on” usage fee per Concurrent Access User embedded in the SaaS pricing model I am proposing. I know this does not release funds for you now but it does create an Annuity Revenue Stream. I could also consider an initial Licence fee for the OpenEHR Speech to Text (S2T) value-add extension.

I don’t know how this would sit with your funding approach and principles. It is a suggested route.

A route which is less likely of successful implementation is to use the association I am negotiating with a significant ICT services company to raise funds as part of the JV start-up. In that funding model, I could position OpenEMR as requiring a development investment to fully implement S2T in the core functionality. This investment would be allocated to your foundation for implementing the solution. We would have to examine the checks and balances on the governance aspects to safeguard all the parties from contingent business risks.This would appear to be a much more principled route as the “open” philosophy will not be impacted.

Please provide feedback to assist me in developing a sales strategy. Do you have an idea of the required funding?

I would appreciate getter clarification on the current status of the extensions for S2T in the Chrome environment. Have your efforts at embedding an open-source speech 2 text product been successful?

If not Dragon and Word would be a route that I could with the obvious constraint that Dragon SaaS pricing is not an inhibitor. A big part of my financial modelling has to take into account a $US to ZAR exchange rate of 1:13!

I hope to hear from you soon as I am extremely excited in seeing your work instantiated into OpenEMR as it is a major step forward in “at source” digitisation of consult notes without “keyboard/transcription drag”.


Thank you for the detailed overview.

Do you have an idea of the required funding?

There’s probably about 20-40 hours of work to really configure, test, and fully document the solution. I’m sure we can find some folks in the community that would be willing to list out their hourly rate. Basically, I couldn’t find anyone to do this project on a volunteer basis and I’m too busy with other projects, so sponsorship is likely the best route.

I would appreciate getter clarification on the current status of the extensions for S2T in the Chrome environment. Have your efforts at embedding an open-source speech 2 text product been successful?

The idea is to use Simon, which is open source. However, it needs to be configured, tested, and documented for use in OpenEMR. It will indeed work with Firefox but we need to explore the chrome support.

I don’t know how this would sit with your funding approach and principles. It is a suggested route.

We accept donations and sponsorship and do most of the core feature work as charity of course. The only request is that we keep everything in the open and check it into the codebase for all to use. If you have a really complex project or set of bespoke changes, I would suggest working with the vendors, however, I see this project as something a community member could do with some sponsorship.

Let me know your thoughts. I’m assuming you will want to support the english language for speech 2 text?


1 Like

here is my nickel worth of thought.

We sometimes in our hast to come up with a solution. We overlook the obvious.
If it has been ten years since the last time you used Dragon. I would suggest buying a really cheap version of the program and try it out. ( I don’t work for Dragon )

Speak 2 Text has come a long way in ten years. I use it often when texting on my phone.

In my view, applications like Dragon does not need to be integrated. It is a desktop application that is activated when the user tells it where they want it to type. The end application does not matter.

So, in the case of using a product like Dragon with OpenEMR. It is simple.

Here is what I do, I open the browser on my phone and go to OpenEMR. I open the patient chart where I am making a note. I speak my note and the words magically appear in the text box. ( Look, mom, no hands )

This can be achieved with most tablets be it android or iPad/ios.

1 Like

Thank you for the input. You are correct. I will get a copy of Dragon and try it out.

plain old home version 13 seems to work fine

1 Like

Dragon 15 professional individual works splendidly to dictate into all fields.and it already recognizes drug names and medical terms.


This topic has some new life from the AWS product HealthScribe. I am adding it here for anyone that looks through this thread. As said in this thread, things change in time and in the six years since this thread started. Generative AI has burst on to the scene and can replace Dragon.