The Eight-Thousand Pound Elephant in the Room

By Hugh Via, TranscriptionGear.Com

Okay, don’t look now but there is an Eight-Thousand pound elephant in the room and everybody just seems to be ignoring it in the hope that it will somehow just go away. In this case the Eight-Thousand pound elephant is speech recognition and I am here to tell you it is here to stay. Not only that but it is getting better with each new version and each new more powerful generation of personal computers.

Transcriptionist have survived many technological advances in the past and will survive speech recognition. Each new advance raises the hue and cry that transcriptionist will be thrown out into the street. It began as medical secretaries taking shorthand notes while the doctors dictated. It then evolved to doctors dictating into a machine that recorded the voice on wax cylinders that would then play back the recording for the secretaries to type out on the trusty Royal manual typewriter, sticky keys and messy ribbons and all.

The next breakthrough was putting these recordings on magnetic tape. Big bulky machines that featured a double-sided cassette that could hold 30 minutes of dictation complete with the dictators chewing noises, pencil tapping and other noises that regrettably add little to the quality and listenability of the dictation. Those accompaniments are still with us today along with many new sounds that the portable units pick up in some of the strangest places.

Next came the digital revolution. No longer did we have to fumble with tapes. No more driving to the doctor’s office to retrieve tapes so we could transcribe them. Things we really getting interesting. We could transcribe dictation anywhere we had access to a computer and a phone line. We could transcribe for a doctor located across the country as easily as the doctor down the hall.

The transition from magnetic tape to digital has taken about Fifteen years and is almost complete. I’ve heard there are a couple of holdouts somewhere in the foothills of the Rockies who are still using tapes.

Each of these technological advancements has meant that authors and transcriptionist alike must adapt. Sometimes it is the authors that are first to want to move to a new method but much of the change is brought about by the organizations that employ the doctors and transcriptionist. Cost cutting and improved efficiency are the goals administrators have in mind when new dictation technology is deployed.

Each advancement has made the process faster and more efficient after the initial kinks are worked out. Medical reports can be completed is the shortest turn around time ever. Transporting documents no longer requires couriers or mailing/faxing. With a few keystrokes we can send the document to any destination on our network or and computer with internet access.

So, what about this Eight-Thousand pound elephant? Speech recognition is here and it works, for the most part. It is not perfect but then again neither are any of the humans involved in the dictation/transcription process. There are several ways that speech recognition is being successfully deployed. Front-end, Back-end and Re-speak are three of them.

Front-end speech recognition is where the author dictates to a PC running a speech recognition software application. As the dictation is created the author can make changes and corrections as the dictation progresses. Many younger doctors and technologist like this method because they are comfortable with the computer and have been using them since they were a kid.

This method does not make the best use of the doctor’s time, however. The distraction of watching the text appear on the screen and the temptation to continuously edit the report will reduce overall productivity.

Back-end speech recognition is where the author dictates their report to a recording device and the voice file is then sent to a computer where the post processing software will run the voice file against the speech recognition application and a text file will be generated.

The Back-end speech recognition process does not create a finished report because there will be words that are miss-recognized or gaps in the text where the application could not understand the author. Documents produced by physicians with heavy foreign accents often have a high number of missed words.

The two methods above still require that a transcriptionist (now called an editor) proof read the text file and verify that the text matched the recorded report. This requires a slightly different skill set than outright transcribing. Still, it is essential that the report be a true representation of the spoken report.

Editing a report that has been handled by Front-end dictation processing can be much more tedious than working with a Back-end file because of the authors editing.

Okay, what about the third method? Re-speak is where a transcriptionist listens to the author’s recording and re-speaks the dictation word-for-word into a speech recognition application. This is going to be a great way for transcription services and independent transcriptionist to take advantage of speech recognition. Instead of dreading the implementation of speech recognition you can embrace it and make it work for you.

Many of the stand-alone speech recognition applications like Dragon NaturallySpeaking are licensed by the voice. If you have 100 doctors that you transcribe for you would need a license for each doctor, or voice. If you are an independent transcriptionist with 10 doctors you are still looking at buying 10 voice licenses. It doesn’t take long before you are looking at spending tens of thousands of dollars in software licenses.

With Re-speak the transcriptionist wears a headset as in conventional transcription but also uses a microphone. It take a little getting used to but before long you will be able to move at a pace that is quicker than typing. Speech recognition can handle upwards of 160 words per minute.

It seems that women have a distinct advantage over men in this process. It has something to do with how the female brain processes information. Women can process multiple streams of information simultaneously. So listening and speaking concurrently is much easier for women to learn.

So if you have a single license for a top notch speech recognition application like Dragon NaturallySpeaking Medical for the PC or MacSpeech Medical for the Macintosh will allow you to use speech recognition to create text reports for any number of authors. Author eccentricities will no longer be a problem as you listen and speak. Keyboard use is cut to a minimum as is RSI to the wrists and hands.