Which Version of Dragon Should You Buy?

Someone asked me recently whether their friend should purchase Dragon Home or Dragon Premium.

https://twitter.com/ewaccess/status/522493851878297601

The Short Answer

Get Dragon NaturallySpeaking Premium 13 for Windows. Dragon Home is missing useful features. Most of Dragon Professional’s useful functionality can be achieved using third party tools.

Decisions, Decisions

Expanding the question from Dragon Home vs. Dragon Premium to the whole universe of Dragon-brand computer products, the question of which one to buy is more complicated than it should be. It involves three factors that I’ll consider below:

• Which platform (Dragon NaturallySpeaking for Windows, or Dragon Dictate 3 for Mac)?
• Which version (For Dictate, 3 or 4. For NaturallySpeaking, the latest version is 13)?
• Which edition (Home, Premium, Professional, Legal, or Medical)?

Which Platform?

This decision is the easiest one to make. If you want to run Dragon on Windows, get Dragon NaturallySpeaking. If you want to run Dragon on Mac, do you want to use Mac OS while doing it?

For the Mac

Quite simply, if you’re using a Mac and you want to run a Dragon program designed to run natively on Mac OS, you should get Dragon Dictate for Mac. The current version of Dragon Dictate for Mac is version 4. Personally, I’m still using version 3. I wrote a Dragon Dictate 3 review in 2012. Truth is, not much has changed about Dragon Dictate since then. The core speech engine (which recognizes your voice and transcribes it to text) is better with each version, but in terms of new features and capability from its predecessor, Dragon Dictate 4 was such an incremental upgrade over its predecessor that I wasn’t compelled to write about it. My Dictate 3 review stands.

Dragon Dictate for Mac isn’t the only option for Dragon users on Mac hardware. You can install Dragon NaturallySpeaking (the Windows version of Dragon) on Windows, provided you can install Windows on an Intel-based Mac. If you go this route, you can either install NaturallySpeaking on Windows using Boot Camp or you can install NaturallySpeaking from within a virtual machine.

Personally, I run Dragon NaturallySpeaking on my MacBook Air using a VirtualBox virtual machine. People have written volumes on whether BootCamp or virtualization is superior. Suffice it to say that Dragon NaturallySpeaking 13 runs very well on my virtual machine (Intel i7 Processor, 4 GB of RAM, 60 GB SSD), but that if I were looking to get a computer specifically to run Dragon NaturallySpeaking then I would get a Windows PC. Virtual machines are convenient for when you occasionally need something from multiple operating systems, but they’re not worth consistent, intense resource usage (by which I mean the whirring fans and toasty lap) that come from hours of dictating on a FrankenMac.

For the PC

If you’re going to get DNS, the question of which version number to get is easy. Always get the latest version that you can afford. Unlike Dragon Dictate, each release of Dragon NaturallySpeaking is faster, less-resource intensive, and more powerful than its predecessors. Dragon NaturallySpeaking 13 is the best version of Dragon there has ever been. The only caveat to the advice is to always get the latest version of Dragon that you can afford is to avoid Dragon NaturallySpeaking 11 or 11.5. Version 11 introduced a new UI and a new speech model. Both of those things were ultimately useful (and they have been carried over and improved upon in subsequent versions), but at the new speech model introduced significant latency that made Dragon borderline unusable for users who weren’t comfortable changing their speech model and modifying the options.ini file.

OK, I want Dragon NaturallySpeaking. Which edition should I get?

You want Dragon NaturallySpeaking 13 Premium. This is true if you are:

• New to Dragon
• A user who is not concerned with full voice control of their computer or automating complex tasks with voice macros
• A web accessibility professional looking for a cheap, fully capable testing tool
• A user with a disability who wants a cheap alternative to Dragon Professional for personal use

But Dragon Home is cheaper! Can’t I just get that?

You could, but you probably shouldn’t. If you like speech computing at all, you’ll quickly become frustrated by its shortcomings. Nuance offers a helpful Feature Comparison Matrix [PDF] that itemizes the differences between Dragon Home and Dragon Premium.

Among some of the things you can do with Premium that you can’t do with Home:

• You can dictate into Microsoft Excel and Microsoft PowerPoint, and control those programs by voice.
• Advanced Correction plays back what you said (in your own voice) while you edit your document. Knowing exactly what sounds Dragon transcribed is essential when trying to decide how to effectively correct any mistakes it makes.
• You can use Dragon with a mobile app for Android or iOS in lieu of a microphone
• You can save and import your custom vocabulary in case you want to carry it over into a new profile, a new install or new version of Dragon
• You can create simple “text-and-graphics” commands, such as “my logo” to insert a logo a document, “my signature” to insert an e-mail signature, or simple phrases to insert often-repeated text.

Advanced Correction (the ability to hear what you dictated when you’re trying to figure out whether Dragon mistranscribed something or whether you misspoke) is a can’t-miss feature, and that alone is worth the price of admission for Premium.

It’s a KnowBrainer

The biggest reason to prefer Dragon Premium to Home is that Premium is extensible. Premium is compatible with third-party add-ons like Vocola and KnowBrainer. With these, you can use Dragon Premium to create custom voice commands to automate any task on a Windows PC.

I have also used Voice Computer, SpeechStart+, and ShowNumbers Plus. All of these tools offer additional hands-free functionality at a cost substantially lower than purchasing a Professional version of Dragon, and none of them are compatible with Dragon Home.

My favorite add-on is KnowBrainer. Not only does it work with Dragon Premium to offer advanced scripting functionality comparable to that of Dragon Professional, but it is free to people with disabilities and assistive technology professionals.

Why should I get Dragon Medical?

• If you or your practice need to be HIPAA compliant
• If you need to dictate into an EMR on Nuance’s EMR Whitelist
  If you need to use Hidden Mode [Flash demo]. Hidden Mode is an exclusive feature of Dragon Medical that allows you to dictate into a non-active window.
• If you want to save time by not having to create your own specialized medical vocabulary.

Dragon Medical also contains every feature of Dragon Premium and Professional.

Why should I get Dragon Legal?

Unlike Dragon Medical, I don’t see the value proposition of Dragon Legal. Dragon Legal contains no extra features or functionality beyond those of Dragon Professional. Dragon Legal does have an extensive legal vocabulary, but the Professional and Premium versions of Dragon can transcribe most legal terms and styles accurately by default. I think that most Legal professionals will be well-served by buying Dragon NaturallySpeaking Professional and simply adding any unknown words or phrases to its vocabulary. Moreover, Dragon Legal’s language model is for American lawyers only, so if you’re outside of the United States then its value proposition is further reduced.

What about Dragon Professional?

I use Dragon NaturallySpeaking 13 Professional. For years I was satisfied with my combination of Dragon Premium and KnowBrainer, which gave me full, hands-free access to my computer and the ability to create custom commands for $99. I recently outgrew it though and upgraded to using Dragon Professional even for personal computing. Dragon Professional has every feature in Premium, and some additional ones that I love:

• Deferred correction. I can save a copy of my spoken dictation with my documents so that I can go back to them over multiple sittings and still remember what I said. This performs identically to the Advanced Correction in Dragon Premium, but with Premium you lose your spoken when you close a document, meaning you need to edit your dictations in the same sitting that you compose them.
• Step-by-Step commands. Dragon Professional contains a GUI that lets you build custom voice commands by stringing simple actions together from a list. This is the quickest and easiest way to create custom voice commands.
• Open-ended commands. New to Dragon 13 Professional, it’s now possible to create commands that accept free-form dictation. Meaning, I can create commands to “Search Website X for [whatever query I speak]. Dragon Premium has long had the ability to search Google, Bing, Yahoo, and Wikipedia for free-form queries, but the ability to create custom searches or to act on any file in a specific directory will be a big time saver for me.
• Easier control over custom vocabularies. Dragon Premium lets you import and export custom vocabularies, but Dragon Professional gives you more granular control over custom pronunciations.
• It is compatible with Utter Command, a third-party add-on that enables you to perform complex actions using an efficient, intuitive syntax.
• Dragon’s Advanced Scripting engine. If you don’t have a third-party command utility, Dragon Professional lets you create powerful, flexible macros to perform any action by voice.

Most of these features are for power users. If they appeal to you, Dragon Professional is worth the price of admission.

Correction- October 20, 2014: This article originally implied that Utter Command was compatible with Dragon NaturallySpeaking Premium. This has been edited to clarify that Utter Command is only compatible with Dragon NaturallySpeaking Professional, Medical, and Legal. I regret any confusion.

The Viking and the Lumberjack: Episode 1

I was thrilled to be part of the first episode of The Viking and the Lumberjack, a playful take on digital accessibility featuring my friends Billy Gregory and Karl Groves. For the record: I don’t have a favorite disability, but my favorite terminal condition is humanity.

(My segment starts at 4:18, but stick around for the Androne!)

WordPress 2014 Theme Brings a New Look and Better Accessibility

Appetite for Dictation looks different today. Accessibility was one of my priorities when I launched this blog in 2012. Then I got lazy.

I asked Nacin which WordPress theme was the most accessible (who better to ask?), and he told me to go with Twenty Ten. Among many others, the theme is still used by Marco Zehe, who is on my must-read list (and should be on yours too, if you follow web standards, accessibility, or Mozilla). I’m sure it was good advice at the time, but accessibility is relative.

That’s not a noncommittal statement to mean as “accessibility is entirely subjective” or “accessibility is a moving target.” Accessibility is certainly measurable, repeatable, and achievable. But accessibility is about finding the best way for the most people to use something (be it information, a system, a service, or an infrastructure) as quickly and easily as possible in whatever their current circumstances. As technology, tools, and techniques change, the best way to make something accessible might change along with it.

As this blog aged, I didn’t bother to review its accessibility relative to the other themes that were out there. As such, I was inadvertently excluding potential readers. A friend recently let me know that he tried to comment on my “How Not to Train Your Dragon” post, but he couldn’t. That’s on me.

The WordPress accessibility team has demonstrated a commitment to accessibility. As part of that, people who create WordPress themes can choose to have an accessibility audit conducted by the WordPress team. Those that pass the audit get the Accessibility-Ready tag. The Twenty Fourteen theme is Accessibility-Ready, and it was a snap to set up.

I hope this is easier for everyone to use.

Accessibility DC: Assistive Technology Users and UX Testing on July 16

Thanks to everyone for coming to the last Accessibility DC event. Special thanks to Devorah for her helpful and hilarious accessibility toolkit presentation. Thanks are also due to everyone who trekked downtown for their flexibility as we adjust to the new cadence and logistics with the new conference space.

Despite any headaches, though, the new conference room is totally worth it. Giant monitors on every wall make every presentation feel like it’s being delivered in a planetarium — and yet, the space is intimate enough that people can talk to each other and work together.

Join us for the next meeting on Wednesday, July 16th from 6:30-8:30. As always, registration is free and open to the public but you MUST register on EventBrite if you’d like to attend.

We don’t have a formal presentation planned, but we want to start a discussion on including assistive technology users in usability testing. How do you fit in? If you’re a designer, a developer, a tester, a project manager, an assistive technology user, a UX professional, or are just curious about the intersection of people and technology — I want to talk to you.

We’re hoping that this discussion will guide some more formal planning and outreach to the wider DC UX community. Hopefully we can get some people with some experience in usability studies and testing (maybe even those with products they’d like tested) together.

Meanwhile, I know a guy who can do some Dragon testing.

If you’re in DC (or plan to be in the fall) and want to get involved with AT and UX, please feel free to reach out to John F. Croston III or myself.

Accessibility DC: “Quality and the Secret Mission” on June 25th

Accessibility DC is back and we have a new home! After a great five years hosting monthly(-ish) meet ups at the Martin Luther King, Jr. Memorial Library, we’ve moved to the McPherson Building in downtown DC.

I’m so excited that Devorah Firestone will be christening the new space with her talk Quality and the Secret Mission (Slides subject to change — but they’re already awesome).

The McPherson building is at 901 15th St. Northwest, 20005. It’s 400 feet from the Vermont Avenue exit of the McPherson Square station on the blue and orange lines. As always, Accessibility DC is free and open to the public. But, there’s a (small) catch: while we’re still feeling out the space, there’s a hard cap of 30 people who can attend in person. So, in order to get a spot you MUST register on Eventbrite 

(If you’d like help registering, please let us know!)

If 30 people register to attend (we’re over 20 so far!), we’ll put together a waitlist. In the meantime, would you like to contribute to Accessibility DC?

• Volunteer to give a presentation, demonstration, lamentation, sililoquiy or performance piece.
• Ask a question and start a discussion on a digital accessibility topic that’s important to you.
• Do you have any ideas for projects we should take on?
• Do  you know anyone who’d like to talk to us?
• Do you have a bigger (metro-accessible, please!) space you can lend us once a month?

Feel free to talk to me or our Head Cat Wrangler, John F. Croston III.

Thanks for your interest, your participation. and your feedback. I really appreciate it.

Slides: Dragon Sucks! Digital Accessibility for the Physically Challenged

How (Not) To Train Your Dragon

Whenever I speak about Dragon NaturallySpeaking as a web accessibility testing tool, I take care to remind the audience of three things:

• Dragon Premium is a complete testing tool. The Professional version isn’t necessary, and the lower versions (i.e., Dragon Basics and Dragon Home) lack essential functionality.
• If you pay more than 99 USD (+tax and shipping) for Dragon Premium, you’re paying too much. Edit: A Canadian colleague informs me that the going rate for Dragon Premium (English) in Canada is $129, which appears to be the current price on Amazon.ca.
• Spending more than five minutes creating and training a Dragon profile isn’t necessary.

I mention these three things in an effort to remove the most common objections to using Dragon as a testing tool in the first place: using it does not require a big investment in terms of time or money. Regarding money: there’s an argument to be made that Dragon Premium (and comparable voice-driven interfaces) should be considered “nominal cost” assistive technology for the purpose of CVAA compliance, but that argument is outside the scope of this brief post.

Regarding time: why shouldn’t testers be concerned with training a Dragon profile?

When using Dragon to test the accessibility of an interface, the tester should be concerned with two things:

1) Can a user interact with an element on screen by issuing a pronounceable, understandable, voice command?

2) Can a user dictate into text-input areas?

Regarding #2: Verifying that Dragon can dictate into a text-input area either directly (when the DragonBar indicates that such place offers Full Text Control) or indirectly (via the Dictation Box) is all that should be required of the accessibility tester. Verifying that text can be dictated accurately is not the tester’s responsibility; that’s on the end user. The quality of the dictation is determined only by the user and their equipment, and one of those times when Accessibility Is Not Your Problem, as Joe Clark might say.

So if you, as an accessibility or QA tester, don’t have to judge the quality of dictation, why should you spend even a minute helping improve the quality of your dictation? An un-trained Dragon profile recognizes text at over 98% accuracy out of the box, and it recognizes commands at a still superior rate.

Instilling Bad Habits

The argument above for not training Dragon when testing can be seen as cold, so I’ll make a pragmatic argument as well. It’s based on the 98% number above. Simply: most users will never achieve higher than 98% accuracy, and some will dramatically reduce their accuracy during the training process.

Training Dragon NaturallySpeaking and its predecessors used to be a daunting task. Part of this perception is practical: the training process really did used to be slow and time consuming. (It’s not like that anymore. Training a profile in Dragon NaturallySpeaking 9.5  and above can be done in 10 minutes.) Part of this perception is a result of the way Dragon is marketed. The first things perspective users hear about Dragon is that it can be “trained to understand you,” that it “learns from its mistakes” and that it “gets smarter as you use it.”

In my experience training and communicating with fellow Dragon users, I’ve found that this knowledge does more harm than good. When training their profile, people think that there’s so much riding on the training process (“ooh, this is important! It’s the only way the computer will understand me! I’d better not mess up!”) that people tense up during this process and don’t actually speak naturally. They over-enunciate, over-articulate, and use exaggerated volume and pacing. They don’t speak naturally. They train the software by speaking to it exactly how one shouldn’t, and then when they relax (once they’ve gotten past the spectre of the training process), they wonder why Dragon makes so many errors. It’s because the relaxed, naturally-speaking person is not the model NaturallySpeaking was trained to understand!

What if 98% isn’t good enough for me?

When people talk about speech recognition accuracy, we do so in terms of percentages. Dragon is 98% or 99% percent accurate. That sounds impressive! I’d be happy with those scores on an exam. But in reality, that metric reflects that Dragon makes one or two errors per 100 words dictated. When you say it that way, that’s a lot. Needing to correct two errors per 100 words dictated can be a big hit to your productivity, which discourages people from using speech recognition in the first place. In reality, my accuracy is much higher than 98 percent. I can dictate paragraphs and pages before Dragon makes an error. 

So, if training isn’t necessary to getting good accuracy, how do you do it?

1) By using proper dictation technique

2) By using the Vocabulary Editor

3) By using proper correction technique

All three of these things are demonstrably more important to speech recognition accuracy than training. I hope to add more detail in the future.

Should anyone do training, ever?

Training Dragon is not necessary and potentially harmful, but that’s not to say it’s useless. You should train Dragon if

1) You understand that’s it’s only useful if you speak during training like you will during normal usage
2) You plan to use Dragon to dictate text as much or more than you plan to use it to access web and software interfaces using voice commands.

If you’re going to train Dragon, I recommend using the starter text “What to Expect from Speech Recognition” (in Dragon NaturallySpeaking 11 and above). It can be fun to read the sample texts from Dave Berry, Alice In Wonderland, or President Kennedy, but the “What to Expect…” training text is useful in that it explains how Dragon works while you read it.

The training text “Talking to Your Computer” (in Dragon NaturallySpeaking 10.1 and below) was masterful. Simple, useful, and unobtrusive, it is one of the better pieces of instructional design I’ve ever encountered. In prior years I recommended people train Dragon just to read that text. “What to Expect from Speech Recognition” is a little more blunt in its presentation and a little less enjoyable to read, but still functional and worthwhile.

A good discussion on Dragon as Assistive Technology

Every so often I learn invaluable things from the #a11y hashtag on Twitter. This was one of those times. Thanks to everyone who participated in this Twitter discussion on Dragon as Assistive Technology

Abstract: Live, at Accessibility Camp Toronto

I’m excited to be participating in Accessibility Camp Toronto on Saturday, November 16. It will be the first Accessibility Camp I’ve been a part of other than Accessibility Camp DC in my backyard. Accessibility Camps are structured as an “un-conference,” meaning there is no set agenda for the day. The schedule is determined on the spot by the participants, each one of whom tells the room what they’d like to learn about. Admission is usually free or minimal cost, with the only requirement for attending being a willingness to share knowledge. It’s an informal structure that encourages attendees to ask questions, take risks, and put themselves out there. I always leave Accessibility Camps better than when I rolled in.

For the last few years, I’ve avoided giving presentations at Accessibility Camp DC in favor of giving Dragon NaturallySpeaking demos during lunch. This is a win-win situation: there are more speaking slots for people smarter than me; I don’t have to speak for 45 minutes; and anyone not impressed with my demo can’t be too angry because lunch comes with free sandwiches.

The good folks at Accessibility Camp Toronto have asked me to speak this year, which is great. They’ve asked me for an abstract (what kind of madness is this!? For an un-conference!?), but I’m happy to oblige because the organizers are awesome.

I’ll be speaking from my experience as someone with a mobility impairment. I’ll be speaking on:

• the challenges I encounter when trying to access digital information
• some things designers and developers can do to mitigate those challenges
• the assistive technology I use to mitigate those challenges
• my assessment of the “average user.” Why aren’t we more vocal? Why aren’t we better advocates? Doesn’t our AT suck, and why are we OK with that?

I’ll be explaining, demonstrating, or otherwise shamelessly raving about some of the following assistive technology for Windows:

• Dragon NaturallySpeaking
• KnowBrainer 2012
• Windows Speech Recognition
•  Utter Command
• SpeechStart+ 
• The Kensington Expert Mouse
• NaturalPoint SmartNav (commercial demo)

I could also talk about access to Mac OS or mobile content, but I get the impression that most of the audience is curious about business productivity, where Windows still reigns supreme. That said, there’s another advantage to an un-conference: the talk will be driven by questions and whatever you all want to know.

If you see me in Toronto, come say hi!

A Brief Followup on the WordPress Admin Video

The most interesting sequence for me is what happens at the 41-second mark.

WordPress Admin after saying “click text”

When I say “click text,” the only thing I can see is a flag with the number 4 by the “Inspire Me” text link. If the first visible item is the fourth one tagged, that means there are 3 items that I can’t see. There’s a couple Dragon commands I can use to discover each of those items even though I can’t see them.

Those commands are Next Match and Previous Match to cycle between the numbered elements, and Click That to activate the selected, numbered element. This process is slower than directly calling “click 4″ or “click 19,” but it allows me to discover blindly what 1, 2 and 3 point to when they’re obscured by page design.

1. The first element with “text” in its label or alt text is “New Post.”
2. The second element is “Notifications”
3. The third element is “Photo”
4. The fourth element, as I can see, is “Inspire Me”

When I issue the command “click text,” Dragon will number either text links or HTML elements with the word “text” in their label or alternative text.

Oddly, it appears here that the “Photo” button has something describing it as “text” whereas the “Text” button does not.