Pages

Thursday, August 25, 2016

Can You Hear Me? Can You Hear Me? Comments on VR

By I,, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=2225868
It is presumed that voice recognition software will replace the keyboard and the manual operating systems of all the world's computers in the future. Well, they have also been predicting flying cars for about 75 years or so and we are really no closer to that reality than we were back in 1940 when this photo was taken.

By Kobel Feature Photos (Frankfort, Indiana) / State Archives of Florida, Florida Memory, Public Domain, https://commons.wikimedia.org/w/index.php?curid=30423797
 The problem with flying cars are similar to those with voice recognition. You can make one, but what you really need is a whole infrastructure of support and a system where the product can be used effectively. You might think that VR or voice recognition is a panacea for those who cannot adequately use a keyboard or a tracking device, i.e. mouse or trackpad. But in reality, the keyboard has one thing that VR has not and that is the ability to make quick and accurate edits of typos.

Even if VR becomes so ubiquitous that it replaces the manual controls of common appliances, the range of commands needed to operate, say a microwave oven, are so simple and limited as to make the problem somewhat trivial. Genealogy is not simple and what we type and record is far from the directions to heat some soup for 30 seconds.

As I have written in the past, I have been using VR for years off and on. There have been tremendous advances made in the accuracy and utility of the products, but for genealogists, we are just about at the very beginning of the development. Let me demonstrate the problem. Here is a rather simple quote from my family tree.
Name
Thomas Parkinson 
Sex
Male 
Birth
12 December 1830
Farcet, Huntingdonshire, England, United Kingdom

Christening
12 January 1831
Ramsey, Huntingdonshire, England

Death
3 March 1906
Beaver, Beaver, Utah, United States

Burial
5 March 1906
Mountain View Cemetery, Beaver, Beaver, Utah, United States
Not complicated at all, is it? Now, using a very sophisticated VR program, without making any corrections, here is what I get:
Name
Thomas Parkinson 
Sex
Mail 
Earth
12 December 1830
Carson, Huntingdon Shire, England, United Kingdom 
Christening
12 January 1831
Ramsey, Huntington Shire, England 
Death
3 March 1906
Fever, Beaver, Utah, United States 
Burial
5 March 1906 
Mountain View Cemetery, beaver, beaver, Utah, United States
Now, there are commands that could resolve a few of these issues, but the reality is that I can accurately type the entire entry in much less time, with a higher degree of accuracy than I am willing to spend trying to get the VR program to make all the adjustments and correct all the bad entries. It does not really help me to go back to the keyboard and try to correct the entries. The reality is that while I am typing, I am making a lot of mistakes. Most of those I can correct with one or two keystrokes. But with VR, I am forced to use a whole bevy of commands, most of which will end up making it ever harder to correct the final product. 

If I were simply writing a letter or an email, I could use the VR program and probable get as close as I needed to with only a few minor corrections, but that is not what I do all day. My operation of the computer involves a highly complex set of instructions that include a lot of clicking and dragging items from one place on the screen to another. To give oral commands to do something as simple as dragging and dropping an image and then formatting it, would require many commands and my frustration level would be enormous.

Even if I had a quick and easy way to use VR to move from field to field in a genealogy program, how long would it take me to train the program correctly for every place name, i.e. changing from Huntingdon Shire to Huntingdonshire? As it is, I have an extra line feed in the second Burial entry above, that I cannot get rid of using the keyboard, how could I do the same thing with oral commands if I cannot do it with my keyboard and trackpad? To correct that formatting issue, I have to go into the HTML and edit it directly.

Some years ago, a friend of mine had a car that gave audible, voice warnings. One particularly annoying warning said, "Your door is ajar." Of course, every time the car said that, we both said, "The door isn't a jar, it is a door." But you can begin to see the problem. Language usage is highly complex and even if computers get to the point of functioning like they do in some movies, they will still be annoying at times and blatantly wrong at other times, just as humans are.

VR is a wonderful tool but we have to realize that just because something is useful in one way or another does not mean that it the universal replacement for everything. I recall the scene in Star Trek IV: The Voyage Home when Scotty is confronted with an old Macintosh computer. He talks to it and of course it doesn't respond, so he says "The keyboard, how quaint." But I am guessing that he would have had a very difficult time entering some complex commands solely using VR as demonstrated by the furious typing that ensues in the movie. By the way, the old Mac would not have the computer power to process the commands that Scotty was trying to enter.

In making these observations, I am not disparaging VR. Here is the last paragraph of my post, entirely using VR.

As genealogist, we need to be open to new technology and adapted [adapt it] to our working methodology. We also need to realize, that not all new technology translates into advantages for accomplishing our genealogical goals. Voice recognition is a powerful tool but it is not quite ready to take over the entire child [field] of interfacing with a computer.

Now, after dictating that paragraph, I went back and made the corrections which are shown in brackets and in red. The errors were real words and not caught by the spell checking capability of my computer. When the spell checker or, in this case, VR substitutes real words, making the corrections much more difficult to detect the typos and make the corrections.

My last note. VR usually refers to "voice recognition." But recently, it is coming to more commonly refer to "virtual reality." Even this type of confusion makes using both types of VR difficult. 

2 comments:

  1. Boy, this topic hit a nerve! I've been a medical transcriptionist for nearly 20 years and have witnessed VR come onto the scene. It was supposed to be the be-all, end-all for physicians, save so much time and money! If anything, it takes more of the doctor's time and a transcriptionist has to edit so many errors in a document that they lose time. The only ones making any money are the VR vendors. For home use, with only one voice, the software is relatively okay. For an acute care hospital with over 500 physicians in affiliated clinics as well as the hospital itself it is a complete and utter disaster. If you don't believe me, obtain a copy of your medical records, which is your right, and review them for errors. Most likely, the document has been edited by an Indian transcriptionist and then reviewed by a US transcriptionist before being uploaded to your electronic medical record. If you do choose to use VR for your own personal use, please do a quick proofread. Another place I've noticed bad VR being used is on news websites where a reporter dictates their story and uploads it to the news site. No one proof reads, and the errors are awful. If you're on Twitter, check out DocGrumpy. He uses Dragon Naturally Speaking and even with him as the only dictator it still leaves error and he tweets them. Thanks for bringing this technology to light, Mr. Tanner. Hopefully someday, it will work wonders for us all!

    ReplyDelete
    Replies
    1. That is an interesting insight. I have seen the technology go from useless to useful, but it is certainly not to the point where it can be relied on for crucial operations.

      Delete