Speech Wreck!

Last week, my company embarked on a journey to speech recognition and I am ready to tear out my hair.  I posted this link What The? and some bits out of the manual that were disturbing to me.  Many of you agreed.  We embarked on a bash fest about speech wreck.  It seems like speech recognition is the demise of our craft.  riot

Riot, right?

Not so fast.

Sherry Doggett is a past president of AHDI (2011-2012).  Sherry saw the post and contacted me.  We just had a really great conversation and she gave me some great insight about SPEECH WRECK.  I enjoyed talking to Sherry because she understands why many of us are distraught over the industry, yet, she knows another side.  So, while I came along to prompt the voices of the people, it is important for us to learn from the leaders in the industry.

Sherry is retired as the UC Health Director of Corporate Transcription.  She lived and worked in Cincinnati her entire life and recently retired and moved to the Washington, DC area.  Sherry's transcription department began using speech recognition tools as far back as 1999.  She particularly loves the product to which I am currently struggling with, Dolbey Fusion.  Maybe the best on the market, she thinks.  As the workforce, we are usually not part of planning discussions and we are left to make up our own stories.  All we know is what is handed to us.  We form opinions.  We think it's the truth.  Today, the biggest lesson I got out of this chat with Sherry is that the technology of speech recognition is really a viable and potentially valuable tool, but its implementation process is key.  It can either be successful or disastrous.  It requires an organized rollout, pre-training of the speech engine BEFORE presenting it to MTs, continued YEARS (years, people) of speech-engine training.

The troubling text that we all went kookoo over in the excerpt I presented are vendor comments about how the product works.  Clients must make important decisions about how they want their reports transcribed, and it is key that decision makers work closely with the vendor to make those things happen.

Having a new appreciation and understanding for speech recognition then, the success of it is reliant on a combination of good product, good management, good IT interface, good rollout, good ongoing reviews and followup.

Finally, I KNEW I liked this lady when we met in Orlando, again during our phone conversation.  She gets it.   Sherry's former employer, UC Health, is a Six Sigma organization.  Six Sigma utilizes tools and strategies for process improvement.  As a former quality improvement specialist, I love this stuff!

Sherry is going to weigh in sometime in the next day or so with some more input.  Thank you, Sherry!




  1. Name Hidden Due to Privacy - December 3, 2013, 7:40 pm

    Amen, Linda. I agree entirely with you, and Sherry is a rare awesome manager. I worked 90 hours last pay period and got a check for $530. I’m beat.

  2. Name Hidden Due to Privacy - December 3, 2013, 7:29 pm

    “It can either be successful or disastrous. It requires an organized rollout, pre-training of the speech engine BEFORE presenting it to MTs, continued YEARS (years, people) of speech-engine training.”
    “Having a new appreciation and understanding for speech recognition then, the success of it is reliant on a combination of good product, good management, good IT interface, good rollout, good ongoing reviews and followup.”

    See this is the problem that I have, Debbie. The success of speech recognition is dependent on too many moving parts, good this and good that and good this and that. When straight transcription is mostly dependent on a GOOD transcriptionist, and she would not require (or she already has) years of training to produce a quality document. Sherry IS wonderful and her organization was very fortunate to have her through this process. A lot of facilities are not as fortunate, and the results are at the expense of the MT. .

  3. Name Hidden Due to Privacy - November 27, 2013, 10:02 am

    Found this on the AHDI site. It’s rather a position statement about speech technology with an accompanying article written by Brenda Hurley. I’m linking to both here. What I hope we can start doing is using these useful position statements to inform the decision in our workplaces of what the industry is recommending. I’m going to send these links to my company owners now. They are fixing to start hacking at our line rate before we are anywhere close to increasing our speed with a new system.



  4. Name Hidden Due to Privacy - November 21, 2013, 1:18 pm

    amen. 🙂

  5. Name Hidden Due to Privacy - November 21, 2013, 1:17 pm

    I guess another thing I am hoping our industry leaders would weigh in on, back to my original post on Facebook, how do we reconcile the issue of learning The Book of Style, believing it has some importance, being QA’d over it for years, and then, with the flick of a switch, why is none of that important anymore? VR doesn’t just represent a new tool, it advocates an entire philosophical departure of the previously held important ways in which we document in medical records. Yesterday, something was so important. Today, it’s not. No discussion of moving to a new way of thinking. Just,. boom. VR is in the house!

  6. Name Hidden Due to Privacy - November 21, 2013, 1:07 pm

    I have some behind-the-scenes experience from when I was the training editor at Transcend. I had to work closely with M*Modal each week to try and bring to them the problems being faced by the MTs. I can tell you that they almost NEVER fixed any of the problems. These problems consisted in part of a word or words that somehow got into the system and would pop up randomly throughout reports, across regions and hospitals, for no reason whatsoever. I was told this would take up to 3 months to fix. What??? And it wound up never being fixed. Basically every problem I ever brought to them was turned around and blamed on the MTs. It got to a point where they no longer wanted to deal with me, I was ‘bullying” them, and eventually wound up losing my job over this. These big companies I think really do not care what the MT has to deal with, they just want the work done, and at that they can then pick you apart to death in QA. One of my big things I taught was to not over-edit, the “a”, “the”, etc., but then was told to back off of this subject and now points are being taken off in QA for adding or not adding these things, as well as punctuation. Even Lynn Kosegi wrote an article regarding over-editing a few years back. IF the companies (and we know whom I speak of) would genuinely take an interest in these drafts instead of just paying attention to their bottom lines, MAYBE, just MAYBE, it would be a better world for us!

  7. Name Hidden Due to Privacy - November 21, 2013, 10:11 am

    Our company is launching Voicebrook in two months. MTs will not have editing privileges so what the doctor’s put in is what will come out. It should be very entertaining.

  8. Name Hidden Due to Privacy - November 21, 2013, 7:45 am

    Laura, look in the ABOUT section of the Facebook page. There is a link there. Click on the link and you will find an MTSO document. There you will find a list of companies and information I have gathered about them. All I can say is send lots of resumes out and keep after it. If you want to email me at debbie.schwarm@comcast.net, I will have a look at your resume to see if I can help with it.

  9. Name Hidden Due to Privacy - November 21, 2013, 5:59 am

    So how do I get in on this editing. I am a relatively new transcriptionist (18 months after 2 years of college) and my job for an orthopedic office has been replaced by an EHR system. I have applied to a great many places. The most common response is that I do not have enough acute care experience and my education was too long ago. I have recently interviewed with the local Wendy’s and McDonald’s as I am now behind in all of my bills, some by two months. I think I would be a great editor as I love to read, and read very quickly. But again, I do not have any recent experience as an editor and I do not know where to seek this kind of employment without it. Any help here would be greatly appreciated.

  10. Name Hidden Due to Privacy - November 20, 2013, 11:42 pm

    Another issue appears to be the QA/QC with the larger MTSOs. If that MT does not put in the “a”, “an”, and “the”, or that nonessential coma or semicolon, then they are penalized in pay and stopped from reaching the bonus tiers. QA/QC should never be punitive but educational. How is one to remember the 100s of rules on 100s of accounts to perfect the VR if there is no consistency? The only place that seems to be working at all is with small accounts or in-house employees.

    Sherry, how many doctors are there in your facility assuming it is fairly viable system and how many are ESL? What is the MT to doctor ratio? Are they “pooled” or assigned work? If they are pooled was it after VR was taught using only small increments of your MT team and best doctors or did you pool work during the “teaching of VR?” These things make a huge difference in the initial effectiveness of VR.

    It seems to be so scattered even in the same facility/MTSO with every hospital/clinic having their own requirements that do not coincide with teaching the VR, and does not follow that list of rules for any requirement to teach VR such as Debbie listed, and as many MTs complain QA/QC seems to be more of a tool used often only to keep them from being paid a decent wage. Many once confident and skilled MTs are afraid to correct it, afraid to add something essential, and afraid to question it or send it to QA for the very above reasons. It kills their morale, questioning everything they know. I cannot tell you how many I have heard from that feel worthless because of this insanity, highly educated, years of experience, experts in their field and all they want is out, in frustration and bankruptcy, many fleeing even to a QuickKing or Walmart where they have less headaches and make more money. The practices of these MTSOs need to be across the board standardized. You would think they themselves would want to do this as it makes them more efficient and the MT more efficient; the perfect world where VR might actually learn, haaaaa. Yet, that is not the case. They are not in it for the long run.

    I have no problem typing it the way “you want it”, but I do have a problem with typing it different for every company, hospital, clinic, doctor, manager, and whomever happens to be on QA/QC somewhere out there in “Never Land” including overseas! At this point, I vote with Debbie that it should be a fairly assessed base hourly wage with a production bonus for those that excel. No more questionable and inaccurate QA/QC docking your pay – if you don’t pass probation or fail a fair QA/QC more than once, then heck fire them. Isn’t that what they do in any other profession if you can’t do the job?

    This would stop a lot of the skirting the laws, the VBC only accounts, not paying for headers, footers, enter key strokes; MT might actually be reimbursed for the worst dictators whom delete half of what they say, start over, and then do it again; forcing them to sit during mandatory scheduled shifts with no work and then either take the menial PTO time or make it up off your shift; hiring as IC yet demanding control over everything an IC does; changing character pitch count, not paying for pulled in formats – your own or otherwise; trying to pay you by the word, document, hour of dictation; changing wages, changing of time cards, never paying for OT, nonpayment for hours of research, the 100s of ways they manipulate your pay; the list is endless in this profession and certainly unique to it. Cripes, you have to be a rocket scientist to figure out if you were paid correctly. Perhaps, then they would concentrate more on actually making it work than how much money they can make by manipulating the MT’s wage.

    The running you through constant changes in accounts is crazy; teach the MT, leave him/her there, he/she is best at it. I understand that a certain amount of “flex” with these companies I necessary, but it has gone way beyond that. It is not just their extensive knowledge and skill that makes the MT fly. It is rare at this point for an MT to do 500-600 lph even on VR, but it is the familiarity with those docs and good management that makes them able to do this. Ask anyone one with a 2-year-old child that is learning to talk; my sons had their own special baby speak like all children. People would say, “What did he just say.” I would say, “He said he wants a glass of water or he wants his toy.” How do I know that, familiarity with him and consistency; I hear him all the time, they don’t, and no one is hearing us either.

    I for one never type “the patient has history of xxx. The patient also has history of xxx. In addition, the patient has history of xxx,” but this is what VR forces you to do, verbatim to teach it. This is part of what makes me fast, and it ends up in the long run being fewer characters to pay for.

    1. XXX
    2. XXX
    3. XXX

    Unless it changes the meaning such as patient has a past – remote history (such as a cancer in remission) which I simply add has a past history of to the numbered statement or if I am unsure of doc’s tense, then verbetim. I don’t type “the patient has had a history of” The only time I type in paragraph form is if it makes no sense any other way and always verbatim in allergies and meds. Yes, on a rare occasion this requires me to go back and reformat, but it is rare that I cannot make it a short form, or in PE doc says: Vital signs: The patient’s blood pressure is xxx/xx while the patient’s pulse is xxx, and the patient’s O2 sat is 98%.

    Burp, and sorry, I forgot to add labs, could you please go up and add labs, sorry, I’m all over the place again, I need some coffee, oh yeah, I forgot to do the PE too, could you insert that under meds, oops I forgot to do meds. I had my favorite scatter brain today. He did the recommendations under the ROS, and then proceeded to do all of the above. At least he apologizes and very clear speaking, not ESL.

    It is always a standard format. Vital signs: blood pressure is xxx/xx, pulse is xxx, O2 sat is 98% unless doc says on arrival to ER then it would be; Vital signs on arrival to the ER: BP was xxx/xx. Every highly trained MT already knows how to short form things and leave out redundancies. VR just beats the good MT to death.

    Feel free to back me up here all you suffering MTSO employees.

  11. Name Hidden Due to Privacy - November 20, 2013, 7:50 pm

    I want to add, again, the biggest tragedy IMO about VR is the MTSO and the hospital’s desire to jump to slashing MT pay so quickly. I am still shaking my head about the presentation by Lynn Kosegi in Orlando where she so thoroughly describes these problems, works for MModal, but tells my conference pal after the session that she has not been entirely successful at making her point even with her own company. If you haven’t seen this presentation, it’s a winner! Lynn describes some of the crazy on this topic too. I urge all of you to at least scan it swiftly. Good stuff. http://mmodal2.files.wordpress.com/2011/05/florida_ahdi_201105141.pdf

  12. Name Hidden Due to Privacy - November 20, 2013, 7:40 pm

    Sherry, your comments just remind me of the reason why we need to dig a little for the answers. I know. I found a few typos yesterday….and was mortified 🙂 DolbEy, for one! Yikes.

  13. Name Hidden Due to Privacy - November 20, 2013, 4:58 pm

    Thanks for giving me the opportunity Debbie to share my story with you. And most importantly, thanks to the MTs (or healthcare documentation specialist) who are having the discussions. Just a bit more background on me…prior to becoming a manager in the late 80s, I worked as a transcriptionist off and on for 10 years, while we were raising our three sons. So that being said, I do understand and appreciate. Speech recognition is a technology tool, no more and no less. Technology can be friend or foe, but what I learned from reporting to our CIO is, it is only as good as how the tool is implemented, staff is trained and most critically how process improvement takes place. I am glad to hear some of you report there are success stories in hospital settings like ours at UC Health, but how do you handle what is happening at times in the MTSO world. I don’t have the answer. I do hear from my MTSO management colleagues though that hospital customers insist that 100% (or near 100% of their work) should be done with background voice recognition to drive down costs. Well we know that is not possible. Some dictator profiles will not rise to the level needed for competent editing and just need to be traditionally transcribed. But in the real world, the MT receives the job, deletes what is not acceptable to edit and transcribes traditionally. But the question begs, are they paid the editing rate? In some cases, probably and in others probably not. There must be an honest dialogue between the hospital customer and the MTSO vendor about what is possible with back-end speech and what is not.
    The skills required to edit are very different than traditional transcription. To move through the document effectively and efficiently, specialized training is required. I hear over and over again this is not happening in some places. We found that several rounds of training and retraining were necessary to provide the staff proficiency required to really feel comfortable. We even offered peer-to-peer training. Every staff person had their own tips and tricks to navigate and sharing was critical for team success. I often hear complaints, “I was trained in the beginning and that was it.” Ongoing training is necessary to evolve.
    Now let’s talk about medical content and back-end speech recognition. To implement back-end speech some standardization of style and formatting is necessary to improve the engine. Does that mean one must give away quality? No it does not. But it does change some of the dynamics. We as MTs (and me included) have to let go of every comma, words that have no impact such as “a”, “an”, “the” and become more concerned ensuring that the speech engine is correct in the area of the medical content. I know I am preaching to the choir here LOL. I just want to ensure everyone that in no way implementation of back-end speech means less quality.
    Another positive about back-end speech, I know several MTs who have not had to give up their careers due to carpal tunnel injuries and injured hands and fingers since they could move to editing. One of our MTs sliced her fingers while carving a Thanksgiving turkey and had to have emergency surgery. She was panic-stricken. We assured her that with back-end speech she would continue to have a career.
    Let’s continue the dialogue.
    Respectfully, Sherry Doggett
    PS You wouldn’t believe how many times I spell-checked given MTs are reading this!!!

  14. Name Hidden Due to Privacy - November 20, 2013, 11:16 am

    I’m having EXACTLY the same questions as I am doing this, Monica. As I asked Sherry yesterday, when I get a crappy draft, I just scrap the sucker and do my thing, load my own stuff from InstantText. It’s way quicker, but it’s not helping the training process. And everyone else is doing similar things. So I himmed and hawed overnight about that too and thought exactly what you are saying. How can this work in the environments you describe?

  15. Name Hidden Due to Privacy - November 20, 2013, 10:35 am

    Okay, I guess my question would be how do you make this work with the huge MTSO’s with 1000s of accounts, 1000s of doctors, and 1000s of MTs who are often jumped around on multiple accounts never getting familiarized with the dictators or accounts? These companies to my understanding refuse to take the impossible dictators off the system. I do know a few MTs that work for hospitals using VR that are happy with it, but as stated above, their VR systems have been in place for years and they are not doing 1000s of doctors. They also are being paid a fair wage which no doubt helps. Even if VR was perfection everywhere, if you are only making a pittance a day, I think they still would not be happy with it.

  16. Name Hidden Due to Privacy - November 20, 2013, 10:06 am

    Sherry is spot-on with regards to training of the speech engine. I believe that is the piece missing because it takes time to do that and for probably many reasons that is just not being done. One thing I have wondered about, with our platform there will be some problem and it loses it’s memory so all the voices need to be retrained, very frustrating and makes me wonder if they don’t do a backup of the software?

  17. Name Hidden Due to Privacy - November 19, 2013, 9:05 pm

    We definitely need to talk about management and how the management or lack thereof can make all the difference to an MT and their ability to make a livable income. How can we hold management accountable?

  18. Name Hidden Due to Privacy - November 19, 2013, 7:50 pm

    Appreciate your comments, Rhonda. We need to hear much more of this side of the coin.

  19. Name Hidden Due to Privacy - November 19, 2013, 7:50 pm

    Thank YOU for taking the time to tell us there is hope 🙂

  20. Name Hidden Due to Privacy - November 19, 2013, 7:42 pm

    My company has been using Dolbey Fusion for about 5 years now, and I think it’s a great product, both for SR and for ST. It does require the attention of management, though, in order to make sure that suitable dictators are targeted for SR and that the system is supported so that the SR engine will do the best job.

    I am very productive on this platform and regularly transcribe at 230 vbc lines per hour and edit at well over 400 and occasionally as high as 650 per hour. If your company and management are paying attention, this is a very good platform that can lead to high productivity.

  21. Name Hidden Due to Privacy - November 19, 2013, 7:37 pm

    Thanks Debbie for your kind comments. Just one clarification, we started working with back end speech technology in 1999. Thanks for taking the time from your busy schedule to speak with me today.


Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>