Voice Recognition and Dyslexia

I was just doing some research on voice recognition software and found some exciting links regarding the use of voice recognition with children who have dyslexia. In one study by Martin Miles, Di Martin & Jim Owen it was found that when 12 students with dyslexia were allowed to use voice recognition software for 10 weeks, their reading age increased by 13.4 months, their spelling age increased by 6.1 months, and their written output increased by 45%. Additionally, the students demonstrated improvements in their self-esteem and motivation, and tended to produce more and better quality work. In the research, all experts agreed that Dragon Naturally Speaking was the most reliable and accurate voice recognition software available. In the past, there have been other voice recognition programs, but they have not kept up with Dragon NS as far as reliability.

Dragon Disappointment

I just tried to install Dragon Naturally Speaking on my home computer and it could not install successfully because I do not have enough RAM on my computer. I purchased my computer in 2003 and it has only 256 MB of RAM--1 GB of RAM is required for installation. Other system requirements for Dragon are a Pentium 4, 2.4 GHz processor, and 800 MB to 2.5 GB free hard disk space if doing a full installation with multiple voice files (650 MB minimum if doing a custom installation with only a few components and one voice file). I have been reading some discussion about Dragon NS--it is very accurate but the voice files are very large and some people have run into problems trying to save such large files. Many of the computers in our district are older and they may not support this program adequately. I probably will be able to install it on my work laptop---will try on Monday.

The Deterioration of Handwriting

In the past week I have been thinking about all of the students who would potentially benefit from using voice recognition due to learning disabilities or fine motor delays. A common discussion among the OTs and myself in my district is how significantly handwriting has deteriorated over the past 10 years or so. Most of the OTs believe that technology is at least partly to blame for the decline in fine motor skills in general as kids just spend too much time in front of a screen and not enough time manipulating toys and other objects. Do we fix a problem that has been caused by technology with more technology? Tough question....plus how do we determine when to intervene with voice recognition? It is not always easy to use voice recognition in a classroom setting without distracting others. Many questions to think about...

Dragon Naturally Speaking 9

Well I received Dragon Naturally Speaking 9 in the mail today--it took less than a week to arrive. The voice recorder that was included (for free) is the kind that can be downloaded onto the computer--even came with it's own software. I have seen them for about $150.00 at Radio Shack which is what I paid for Dragon NS--not a bad deal. It also came with a headset that seems to be the same type as the one that I am using now--I got it out of an old Via Voice box. I am looking forward to giving everything a try.

Microsoft Voice Recognition--1st try with a student

Well, today I tried Microsoft's voice recognition with a student in 7th grade. This student has struggled with illegible handwriting since elementary school and is now barely passing middle school due to behavior issues. I wanted to give voice recognition a try with him to see if it would motivate him to begin to complete classroom work. He was able to complete the training with very little difficulty and was able to read the training text even though I initially had thought that it might be too hard for him. He is at grade level academically, so the reading level for training is probably around the 7th grade level. I made the mistake of recording into my voice file-didn't realize that until we were halfway through the training--and didn't know how to create a new user. Needless to say, the voice recognition wasn't very accurate but regardless, he was impressed and excited about the technology. He is going to try voice recognition at home on his mom's laptop--maybe this will motivate him to start completing work at school.
Since then, I have researched how to create a voice file for a new user. You need to go into the control panel and select "speech", then "voice recognition". Then from this point you can create a new user and train your voice. Although Dragon Naturally Speaking is supposed to be the most accurate, it looks like Microsoft Word does pretty well--plus it is free!

Microsoft Word Voice Recognition--Free is Good!

Earlier today, I spent some time trying out the voice recognition option that comes with Microsoft Word. To access it, you first open a Word document and select "tools", then "speech". The program will then lead you through configuring your microphone and training your speech. The initial training takes about 10 minutes and asks you to read a blurb about voice recognition. The vocabulary seemed to be fairly high (middle to high school level?), and may be difficult for younger students to read. However, I was extremely impressed with the accuracy of this program after only one training session. It seems to use the same voice files as WordQ/SpeakQ (or vice versa), so maybe it had a head start from the training I had done earlier. Nevertheless, it was about 90% accurate and had the additional capabilities of voice commands such as "select all" and "delete". I had to do very little editing. Not bad for a free program! Wow!
I also used it to create a Powerpoint presentation, and to write an email so it works in other applications as well. I would be interested to compare all the programs when being used with students who may not have complete fluency and articulation clarity. I plan to try these programs with students this week. I am still waiting for Dragon Naturally Speaking to arrive in the mail...

2nd Try with WordQ/SpeakQ

I just spent another 45 minutes or so trying out WordQ/SpeakQ. This time I actually opened up the manual (PDF file) and read through areas that I had questions about. They recommend 15 minutes of training before SpeakQ is felt to be accurate enough to use--I had only trained 7 minutes the first time and still had about 75% accuracy. Each training session lasts about 5 minutes so I completed two more sessions before putting the program through its paces in a Word document. I am really getting sold on this program! I could tell that the accuracy had increased with additional training, plus I discovered that you can train individual words as needed (any time) by adding them to the dictionary and recording the pronunciation of the word. After doing this with the problem words (it did not have "blog" in the dictionary, and had difficulty with some other words) I noticed much more accuracy. I also added my email address to the dictionary and recorded the pronunciation as "email address" and it worked! You can also create topic dictionaries so if the student is doing a report on a particular topic, it will include words from that dictionary as well. When using SpeakQ, I prefer the preview option and found that using the mouse to choose the correct phrase rather than hitting the corresponding number was the easiest way to go. It will be awhile before I will use it to speak directly into the document, if at all. The drawback of this program is that you really need keyboarding skills to edit your document as it has limited command features, that are limited to punctuation marks. For example, you can say "period" and it will type "." but you can't say "backspace" and expect it to actually backspace--that you will need to do by hand.
I think that this program would be best for students who are able to do some keyboarding but have poor literacy skills. The fact that the words are read back to you after you say them while appearing in written form would be very reinforcing for students who are just learning to read/write and have a lot to say. I think that the students with developmental delays and learning disabilities could experience success with this program while reinforcing their learning. I am going to give this program another try with some of my students who have some speech errors to see if training individual words will improve the accuracy of the program. I think I will have to add this program to my wish list!

1st try with WordQ/SpeakQ

I have been working with WordQ/SpeakQ for the past half hour. The initial voice training took only 5 minutes but I did a second one just for more accuracy. These programs work on top of any program that you would normally type into, so I worked first in a new Word document. I initially worked with the "speak and select" option which allows you to pick the correct phrase from a list (much like Co-writer) before sending it into the text. I spoke about 5-8 words at one time, but when it had difficulty with a certain word, I spoke 1-3 words at a time to get over the rough spots. It did OK with common words, and was about 75% accurate but had difficulty with other words. For some reason, it really had a hard time with "want". I really like the fact that there are many options with WordQ and Speak Q--you can turn the word prediction off and just use the speech, you can turn the speech recognition off and just use word prediction, or you can just have the text read to you that you just typed. The training was pretty painless as well. When you open any program, WordQ/SpeakQ goes with you. I tried opening up my email account but it could not understand "pamj327", even with many tries. I think that there is an option where you can add custom words and short cuts to the dictionary--I will work on that some more. I have WordQ/SpeakQ on a 30 day trial so will keep working with this program until I figure it out. More later....

Add-ons for Dragon Naturally Speaking

I was just reading my email from the QIAT listserve and learned about some add-ons for Dragon Naturally Speaking. A company called KnowBrainer makes a command module for DNS that increases the capability of Dragon, and basically upgrades the Preferred version of Dragon ($200) to the Professional version ($800) for only $170.
Also, there is another program made by the same company called Uni-Voice, which increases the accuracy of Dragon when using it with young children or those with slurred speech. I will have to check these programs out.

WordQ/SpeakQ

Another program that I would like to look at besides Dragon Naturally Speaking is WordQ/SpeakQ made by Quillsoft which I have played with a little bit using a 30 day trial version. The benefit of WordQ/SpeakQ is that is combines word prediction and a talking word processor with voice recogntion. This makes it easier for younger students to train their voice without a more advanced reading level. Basically, the computer reads the words to the student in a 2-5 word phrase and the student repeats it. Then, when actually using the program, you can choose to have the text go into a window to be previewed before you send it into your document. That way the student can correct errors beforehand. This feature can also be turned off if desired. I did try this program with a couple of 4th grade students with Asperger's last month. They each were frustrated after only 5-10 minutes of training because the program kept asking them to repeat words that it didn't understand...both students have some speech errors that make some words difficult to decipher. The more frustrated these students became, the louder and squeakier their voices were...not to mention that the computer did not respond to the insults hurled at it ("you stupid computer!!") I think that I need to train these students how to record into a microphone in a clear voice before proceeding....maybe after a little more speech therapy, plus some maturity... Nevertheless, I am interested in this program due to the fact that it has many features that Dragon does not.

Dragons and Mics

Whenever someone mentions voice recognition software, Dragon Naturally Speaking (DNS) is usually mentioned in the same sentence. Dragon just came out with version 9, so I just ordered it today from Nuance.com . They currently have a special on the newest version of DNS (version 9) for $149.00 which includes a free digital recorder--can't beat that! I also did some research on microphones since it is essential to have a noise cancelling microphone for optimal success. According to some assistive tech experts on the QIAT listserve (Quality Indicators in Assistive Technology) , the VXI Talk Pro Express is supposed to be an affordable and reliable mic for use with voice recognition. I will wait and see what type of microphone comes with Dragon..

Some background..

I am going to attempt to figure out the different voice recognition programs that are most often used for students who have difficulty with the writing process. It can be a bit overwhelming trying to decide which program to use with a particular student and I hope to sort out some of the differences between programs. Since I am supposed to be one of the assistive technology "experts" in our district, I think that it is about time that I jumped into the topic of voice recognition. I have only used voice recognition software with one student so far in my practice as a school physical therapist and that was quite some time ago. About six years ago, I attempted to teach a 6th grade student with muscular dystrophy how to use Via Voice which I purchased for less than $20 on a clearance rack (it must have been the earliest version). The only computer available was the teacher's computer (desktop) so we worked on training the computer to recognize his voice right there in the classroom--not an optimal situation. Our sessions were at the end of the day and his voice tended to be very soft because of fatigue. The fact that the classroom was quite noisy did not help much. Needless to say, we were not very effective in training the computer although we did get to the point where he could get the computer to type about a paragraph but we had to do so much editing that it hardly seemed worth it. We usually spent a lot of time trying to resist the urge to laugh out loud at some of the errors made by the computer--frustrating and hilarious at the same time. Thankfully, this student is now a successful user of Dragon Naturally Speaking thanks to another therapist in the district who worked on voice recognition with him in high school.