After Hours

10 wishlist items for Dragon Naturally Speaking

For those who do a lot of writing, dictation software like Dragon Naturally Speaking can be a big help. But as Brien Posey observes, there's room for improvement.

As someone who writes for a living, I make heavy use of speech dictation software in an effort to boost my productivity and save wear and tear on my hands. My dictation product of choice is Dragon Naturally Speaking. As great as the product is, however, I do have some wish list items for the next version.

1: The ability to disable individual commands

In all fairness, Dragon Naturally Speaking does allow you to disable some voice commands. But I would like to see a series of check boxes that let you choose exactly which commands you want to enable and which commands you want to disable.

For me, the Select command has proven to be especially troublesome. I can't tell you how many times I have dictated a phrase such as "Right-click on the icon and select the Properties command from the shortcut menu," only to have Dragon recognize the word Select as a command. This causes the previous occurrence of the next word or phrase I dictate to be selected. If I don't catch the problem immediately, the next thing I say will overwrite the selected text.

2: Natural language dictation

Another thing I would like to see in future versions of Dragon Naturally Speaking is natural language dictation. Normally, when I dictate an article such as this one, I have to verbalize punctuation symbols such as periods, commas, and question marks. This is fine for writing articles, but it doesn't work so well for transcription.

Recently, for example, someone asked me to develop a PowerPoint presentation, record audio for each slide, and then create a transcription of my audio. I decided to use Dragon to transcribe the MP3 files I created to go along with my PowerPoint presentation. The problem was that Dragon did not insert any punctuation because I had not spoken punctuation commands when I recorded the presentation. It would be nice if Dragon could use sentence structure to figure out where punctuation should go.

3: A comfortable headset

Okay, this one isn't so much a gripe about the software as the hardware that comes with it. Some of the Dragon Naturally Speaking products include a Plantronics Bluetooth headset. The headset seems to work well enough, but the clip that holds it to your ear can become uncomfortable after a couple of hours. It would be nice if the headset were more comfortable.

4: Easy profile migration

Because I travel almost constantly, I have Dragon Naturally Speaking installed on my computer at home and on my laptop. Dragon includes a mechanism you can use to export a user's profile and then import it into a different machine. I haven't had a chance to try the import/export process with version 12 yet, but in version 11, the process leaves a lot to be desired. Importing my profile into my laptop seems to adapt the laptop to my speech patterns, but my custom vocabulary doesn't transfer over. Words I use all the time in my articles, such as Hyper-V, Fibre Channel, and SAN, are not recognized on my laptop and I have to manually add those words to Dragon.

5: Microphone independence

In Dragon Naturally Speaking version 12, the user profiles seem to be tied to a specific microphone. In a way this is completely understandable, because every microphone has unique sound characteristics. It makes sense to adapt the dictation software to a microphone's characteristics. Unfortunately, Dragon seems to depend on a microphone even for operations that do not involve the use of one.

For instance, l week, I was on the go and didn't have time to write an article that was coming due. I dictated the article into my cell phone on the way to the airport and saved the audio as an MP3 file. Once I got checked in and was waiting for my flight, I got out my laptop and told Dragon to transcribe the file. It would not do the transcription until I plugged in my Bluetooth receiver, even though I was not going to be dictating.

6: More entertaining training

This one is nit-picky, but I wish that the training process were a bit more entertaining. When you first install Dragon Naturally Speaking, you are advised to train the software by reading a specific text document. However, the documents you can choose to read are primarily instructional in nature or they come from books we were all forced to read in school. Would it really hurt to include training material that is a bit more upbeat and fun?

7: Right-click > Add

As I mentioned earlier, some words I use frequently are not in Dragon's default dictionary. You can add such words to the Dragon custom dictionary, but doing so involves clicking on the Dragon Toolbar's Audio option and then choosing Improve Recognition Of Word Or Phrase. It would be much more efficient to be able to right-click on a word within a Word document and have an option on the shortcut menu to add the word to Dragon's custom dictionary.

8: Less literal interpretation

Once trained, Dragon Naturally Speaking does a good job of accurately recognizing spoken words. Every once in a while, however, it takes an expression and interprets it a bit too literally. For example, last week I was writing an article and spoke the phrase "a one in a million chance." Dragon interpreted the phrase as "a 1,000,000 chance."

9: The end of unwanted abbreviations

Along the same lines, I wish that Dragon would not try to abbreviate my spoken words. I recently used the word "minute" in a document. Dragon took it upon itself to abbreviate the word as "mins." This meant that I had to go back and manually correct the resulting text.

Dragon does have a mechanism for controlling formatting. You can somewhat control the use of abbreviations or even numerical formatting. The problem is that the interface is a bit ambiguous. It can be difficult to tell exactly how the changes you make will affect the overall dictation experience.

10: A more intuitive user interface

Finally, I would like to see Dragon Naturally Speaking equipped with a more intuitive UI. The interface is fine for basic dictation and training, but some of the configuration options don't make a lot of sense. I doubt if it would take a lot of effort for Nuance to rename some menu options and expand some dialog boxes to make the interface a bit more user friendly.

Other improvements?

What additional changes would you like to see in Dragon? Share your thoughts with fellow TechRepublic members.

About

Brien Posey is a seven-time Microsoft MVP. He has written thousands of articles and written or contributed to dozens of books on a variety of IT subjects.

13 comments
onenrone
onenrone

Hello, Just so you know I have been a voice recognition software user since days pre-Dragon (Via Voice, L&H, Dragon 1) and if I can add to something from that experience, slightly covered here, possibly your readers will benefit. It's ALL about the microphone. 1.) You can get (as I usually do) 98% accuracy with the proper Microphone. However not just the model, which should be as expensive as you can afford (basically higher cost when it comes to mics equals better quality) there's also a world of tech related to audio that can help; equalization, hiss and pop suppression, headset placement, desk mic placement (what I use) etc.etc. And that's just straight in. When it comes to dictation machines, mp3 off the phone then that's best accomplished with separate training for that input. This is one cause, and purpose, of your frustration. Wanting a single user for each and all input. Figure that out and you win the Nobel Prize before the person who accurately translates Linear A. As tech republic puts it, the takeaway here is that you need to train for each input device, mic, file. I also think this is the cause of many not using voice recognition, the Dragon fed commercials that make everyone think it types when you talk... Oh, it types alright, just not what you are always talking. Which brings me to... It's All about training and correcting the program. Again, You can get (as I usually do) 98% accuracy with the proper Microphone. However it is a STEEP learning curve. A constant correction regimen, and learning a new way of speaking period new paragraph. It's ALL about understanding that it takes a LOT of use, correcting, tweaking, did I say correcting, voice input training, and ... tweaking..and correcting. Last and foremost. For those who say, Well I can type fast, and, cut and paste works and I can't get Dragon to do this or that and many other excuses for not using voice recognition, understand how lucky you are. You're not handicapped, not needing adaptive and assistive methods to communicate, you are fortunate to have the ability to move your hands, coordinate your speech and thought that so many who use voice recognition don't. So hopefully your readers will keep in mind their wizardry when writing may not carry over to all others. Hope this helps. Cheers! NR Washington DC

Robert McA
Robert McA

see how that sentence didn't capitalize? It has a total inability to capitalize. It's like it is totally guessing. 50% of the time it gets it right. 50% of the time and get it wrong. online help is horrible. (Latest fury: how do I change my accent setting? The Windows help topic says that you can do it, but then doesn't give you a link! I tried to find it buried under something, no luck) the customer service is always overseas, and none of the agents seem to use the program themselves and their only job is to pass you off to some tip sheet, which of course doesn't help. Finally, the biggest reason I hate DNS is because, despite all this talk about "training", and all the profile importing that I did the upgrade to version 12, the Dragon on my iPhone works 400% better than my PC!!! Of course this is a general meme for me: AppleWorks, Microsoft doesn't. So maybe it's not Dragon's fault. But if you're not hooked, I would not waste your money and especially your aggravation time. Find an alternative. I don't really have an alternative because I'm a horrible typer (Dragon wanted to write "type or" ha!). So looks like I'll be whipping out my iPhone to dictate. hassle.

m-olson
m-olson

Like Cesarcalifornia, I have been buying Dragon Dictate upgrades for many years. The new version, 12, is a touch better than the previous version but it still needs work. While allegedly cross-platform in operation (its supposed to work in any app), it doesn't always work well in some apps. One thing to try is to dictate using Dragon Pad (just say OPEN DRAGON PAD), Dragon's own "word processor," when it doesn't work in your chosen app. Then just cut and paste the result in your app. Not elegant but it works. Since Dragon Pad is presumably optimized for dictation, it may also clear up many problems people seem to be having (e.g., accents, although there is someplace in the program to select an "accent", I think; I saw somewhere that my "profile" was classified as "midwestern").

nmurtagh
nmurtagh

1. Although reasonably accurate, it sometimes made some awful (or funny) errors that might slip through (unlike typing errors that are usually obvious) and be read by a client. (This is with technical and legal translation). 2. I can type almost as fast as dictate. 3. Upgrades are not cheap (while Win7 voice input is cheap -- if you have Win 7 already :-).

cesarcalifornia
cesarcalifornia

I have purchased Dragon NS since version 9 and I have barely used it. I have version 11.5 now. I have always hoped for good results from it, but did not do well with the program (I have an accent and did much better with the Spanish version) and hoped to have time to add words to it, etc, but it is very time consuming. The program does not allow for multiple versions to be installed. It uninstalls the previous version, which is fine with me. I am not sure if it allows for versions in two or more languages (such as English and Spanish) to be installed at the same time. It seems like Nuance does not provide updates (or provides hardly any) and then sometime later asks you to buy the upgrade. This becomes expensive for some or many. I have not received notifications for updates since I purchased them. Each upgrade could cost $99 or more depending on the version and if it is an upgrade or the full version. It seems like Nuance does not provide updates and keeps selling upgrades. I have not purchased version 12 yet. I had more trouble with version 11.5 than with 9.5 because of the program trying to come up with words for little noises it hears. It became a hassle for me, slowed down my computer to a crawl, and I had to turn off the program every time. Cesar

cesarcalifornia
cesarcalifornia

The program requests that only one user be able to use the program. There is no support for multiple users. It would be nice that the program would be able to handle multiple people talking and hopefully differentiate them: User 1: Hello User 2: Hi User 1: How is everything going? User 2: Fine, thank you. I have hardly used my Dragon NS version 11.5 because when I turn it on the program interprets non word sounds as words. So if there is a noise somewhere then attempts to interpret it as words. Then I end up having a bunch of weird words in my documents. I have to check what is going on, maybe reduce the sensitivity of the microphone. Cesar

paul
paul

this software has bad spelling and bad punctuation....maybe it was because my cell phone was ringing....lol

paul
paul

Your number 1 : Maybe if you change your wording to ' choose' instead of select, or scroll to....and click... One must try to think the easiest way of learning another machine, like teaching a child in the begining....the ear piece, well speaks for itself,...and same with the numbers...surely you could word it better yourself. One just needs to remember, elecrical 'brains' as I call them are trying to work at their fastes speed always, so will make errors and try shorcuts... I have always thought that, yes they make things fast enough to do the work, but why not give them a bit more for reserve, instead of 'just' enough. L:ots of 'Brain' type products that do many things need a reserve space and memeory to move things around....I call it the workshop space...if that space is just big enough, many times bits of other projects will get in the way, but if there was a little more room, plus a little extra memory module just for that area....hmmmmm, I would think alot of more precise 'things' could be worked out. So I think when they finally ealise this 'need' not just for the extra room, so to speak, but also some extra memory, possibly power too, for this area... Anyway, I myself will also try this software...as have tried a few already and also found them more bother than they are worth really. Pretty hard to beat a good secretary ! lol

TsarNikky
TsarNikky

#1. This is a no brainer. Include a list of all commands, each one with a checked check box next to it. Uncheck those commands to ignore. #3. Consider the headset as a teaser--just enough to let you decide if the product is going to work. Intense users will chose their own headset. #4, Structure the profile attributes as a single file that can be exported/imported as needed. #7. Set up the auxiliary dictionary mechanism, such as used in Word. You type in the word and then speak the work. [This could be used to handle the "minute" (meaning "tiny") versus "minute" (meaning length of time) problem. The differences in how they are pronounced would tell Dragon how to handle it.]

Slayer_
Slayer_

Instead of having to make all the corrections. Voice dictation never works for me, the cell phone ones are especially bad.

petermat
petermat

Forget all the things you mentioned (except possibly the first.) You say you get good accuracy, but I don't. I have spent many hours training, I have the recommended good usb mic., but the accuracy is still poor - around 90%. this makes it slower than my poor typing. It now languishes in a drawer.

aiellenon
aiellenon

picked this up (Dragon 12, ***edit, just checked, and I have 11.0 not 12.x) just before my daughter was born, thinking (I work from home) I could feed her (from a bottle) and dictated emails... it added some pretty freaky words into my emails every time she made a noise, and as such have not used it since. I even bought the #1 recommended (non $300) headset for using with Dragon. It works pretty darn good when dictating anything of length, but I agree with most, that it is not accurate enough for shorter dictations and after having to edit the document it can actually take longer. Before someone says it takes a while to "teach it" I spend 3 months doing nothing but training 3 days a week, and I used Dragon 6 (I think it was 6, a few versions before they were bought out) A LOT, as in daily, for writing papers and multiple fiction book drafts. I had several issues with it, but knew them well enough to expect them and how to correct them well enough that I could edit while dictating. The training for existing words helped out immensely, If I remember correctly, the old version let you record multiple different ways of speaking the same word (I could say it slow, fast, mumble, deep voice, not so deep voice, mic close, mic far) and it would then get that word correct 95% of the time, no matter the circumstances. The new version seems to not work the same way, it errors during word training if I try to say it differently and after training does not always recognize the word. I think the most useful update would be "background noise cancellation" Second on my list would be a new wireless tech that uses a MIMO style transmit and receive interface, with reduced latency. (this is something you can learn all about if you research wireless headsets and dictation, although not mentioned in the research I did, it would help to alleviate most of the referenced issues with wireless headsets)

Lost Cause?
Lost Cause?

I bought Dragon a couple of Christmases back. I went through the training and hoped for the best. No joy...I was getting about 30% accuracy at best. I went round and round with Dragon Support. Never did really get any help as to why my accuracy was so low. They had me send tape recordings of my voice. No luck. Windows 7 has dictation built in. There my accuracy is about 80%. Too late to get my money back...