When it comes to speech recognition software, Dragon is pretty much the only game in town. Ever since Dragon Naturally Speaking version 9, the accuracy of voice recognition has been very good.
The upgrade from Dragon Naturally Speaking version 9 to 10 improved the accuracy of voice recognition perceptibly; and even version 11 improved the voice recognition noticeably. But Nuance disabled the use of Dragon Naturally Speaking version 11 in Electronic Medical Record programs. They want to sell, Nuance Dragon Medical Practice Edition 1 or 2, several times more expensive to the healthcare market. And the irony of all this is that despite costing several times more, Dragon Medical 11 (Dragon Medical Practice 1) does not work any better for electronic medical records than Dragon Medical 10. You just feel being ripped off!
Having used Dragon Naturally Speaking version 12 for 4 weeks now, I think version 12 is actually a DOWNGRADE and not an upgrade because of following reasons.
1. SLOWER: I have been spending a lot of time watching the spinning dragon in the correction window on Intel i3 processor 10 Gb RAM computer. Sometimes 6 seconds for a correction.
2. BUGGY: The Dictation Box keeps popping up in a variety of programs even when you are directly dictating into the program and do not want it. For example, you are dictating a document in Microsoft Word and suddenly the Dictation Box pops up and all dictation starts going to the box instead of Microsoft Word document.
3. Many of the old formatting commands do not work, and you have to learn a whole set of formatting commands all over again. Nuance is claiming lot of new formatting commands but so far I have seen very few new usable commands.
4. It does not work for Electronic Medical Record programs. And also functions poorly in browser and Citrix based programs. That is where Dictation Box is useful.
5. It does not work on Remote Desktop programs like RDC, Logmein, RDW, JumpDesktop etc.
6. Keeps crashing in Windows 8 and does not close, when you exit the program. Releasing a program in late 2012, tailgated by Windows 8 release, that does not work well with Windows 8.
But version 12 has a few practical advantages.
1. The repertoire of recording devices has increased.
2. You can add multiple audio sources to your profile, instead of having to create multiple profiles.
3. Tutorial is nice, as of now only text, and no videos.
According to Nuance, the accuracy has improved by 20% but the improvement is hard to perceive in day to day use because Version 12 is slower. Slight improvement in accuracy for a slower program is not a trade off to call an upgrade. Version 12 is supposed to have social media commands, but I have not used those.
Based on my personal experience with microphones and recording devices over 15 years on Dragon Naturally Speaking version 3 to 12, I found the following.
A. USB MICROPHONES: give the best voice recognition. Andrea NC-181 VM USB High Fidelity Monaural USB Computer Headset (P-C1-1022300-1) is good value for money. And if you are dictating long documents or do dictation stop-n-go piecemeal then connection between microphone and computer is solid. Dragon or Plantronics USB headsets that come bundled with this program also give good (but not the best) accuracy, but lack mute and volume controls.
B. BLUETOOTH MICROPHONES: Work fairly well, but not as good as USB microphones. PLANTRONIC Calisto Headset with USB Dongle - 81493-02 is very good. The biggest handicap with wireless microphones is that connection between the microphone and the computer keeps breaking and they do not work well for piecemeal stop-n-go and dictation. Every-time you start dictating after a long pause, you have to reestablish the connection.
C. RECORDING DEVICES:Olympus WS-821 Voice Recorders with 2 GB Built-In-Memory gives excellent accuracy, comes with built in USB port and is very reasonably priced. I use the 4 GB version Olympus WS-822 GMT Voice Recorders with 4 GB Built-In-Memory.Sony ICD-MX20DR9 32MB Flash-Based Digital Voice Recorder &Sony ICD-SX712 Digital Flash Voice Recorder give good accuracy but require USB cable and software. Dragon has an iPhone Recorder app which also works very well and gives decent accuracy on iPhone 5 and you can download the sound files to the computer using bonjour on the computer. For dictation on-the-go, and later transcription on Dragon, these are the way to go.
D. SMARTPHONES AS MICROPHONES: I have not had good experience with iOS devices like iPhone or iPod touch, using as microphones, and the main problem is that their connection with computer keeps breaking off, in middle of dictation, which is very disruptive when you are dictating a long document or pause for 15-30 minutes. Getting a phone call or text in middle of dictation can be troublesome. iOS devices are probably OK for short snippets for Facebook posting, as shown in Demo, but not long document dictation. it is supposed to work with Android devices too, but I have not used it personally. Dragon Recorder app on iPhone 5 works very well as a recording device, however.
If you do a lot of dictation, particularly stop-n-go, like I do, for an hour or more every day, then consider hand-held and table USB microphones. Time wasted in putting a microphone on and off the ear and turning it on and off can add up. Instead of turning microphone on and off use "go to sleep" and "wake up" commands - much quicker.
1). TABLE MICROPHONES: SpeechWare's TableMike's 3-in-1, 6-in-1 and 9-in-1 TableMikes are absolutely phenomenal. SpeechWare 3-in1 TableMike is what I use every day and I have not had used a microphone with better speech recognition accuracy than this. For me, working handsfree increases the productivity.
2). HANDHELD MICROPHONES: Like Philips LFH3500 SpeechMike Premium with Precision Microphone and Push Button Operations and Dictaphone PowerMic II Handheld USB Dictation Microphone (Without Scanner) have good speech recognition accuracy and the advantage of adding programmable buttons for activities that you do repetitively in your work day. Philips LFH3500 SpeechMike Premium has a better speech recognition accuracy, motion sensor that mutes the microphone when you lay it flat, a trackball and the buttons can be programmed to work in multiple applications - like Excel, Quickbooks, etc. but Philips software is very buggy. PowerMic II Microphone is made by Nuance, works only with Dragon Medical, has a trackpoint (like ThinkPad), cannot be used in any other application other than Dragon Medical but needs no extra software, comes with software built-into Dagon Medical. SpeechMike LFH3500 is the other microphone I use every day.
If you are already using Dragon version 10 or 11, then stick with it because you will lose a lot of functionality in version 12 without gaining much in accuracy.
If this is the first time you are buying Dragon and you do not plan to use voice recognition for electronic medical records directly in the EMR program or on remote desktop (like LogMeIn, RDC etc.) then go for version 12. Otherwise, stick with version 10 or 11 or grab Dragon NaturallySpeaking Premium 11, [Old Version] at a much lesser price. Version 11 has excellent accuracy, is fairly fast, and is not buggy but does not work with Electronic Medical Records.
Of course, you can still use Dragon 12 with a recorder and the speech can be transcribed later and entered into the EMR program using cut and paste. That works or people who use an assistant to transcribe and enter data into EMR. Dragon Medical actually comes with an Autotranscribe Agent, which runs in the background and turns voice files uploaded into a folder on the computer directly into the text, unattended. If you use that great tool Autotranscribe, then you may not feel bad paying for Dragon Medical.
Dragon NaturallySpeaking 10 Preferred - Medium Box works in most electronic medical record programs and on less powerful computers but is less accurate than version 11. Dragon NaturallySpeaking 9 Preferred [OLD VERSION] works in all electronic medical record programs that I know off. If not at Amazon, you can find older versions at off the beaten path retailers online, very cheaply. But if you work with electronic health record on a Windows 7 or 8 computer, you have no other option but to buy Dragon Medical Practice 1 or 2. The whole point of Dragon 12 upgrade is really to choke off the use of regular Dragon in electronic medical programs.
on September 17, 2012
Having used Dragon NaturallySpeaking premium 12 for about a month, I must admit that I'm completely baffled. They call this an upgrade?
For starters, this version is incredibly slooooooow. And I mean slooooow. It is slow to launch, and slower to close. Recognition is even slower. The promotional material stated that it was "20% more accurate," but I have yet to discover this unique aspect of this upgrade. The previous version (11.5) is far, far superior and much faster, compared to this clunky dog. And in typical Dragon fashion, technical support is a big run around, especially when you're calling to complain about the new upgrade. No one seems to know anything.
I've found that it no longer allows me direct access to dictate my godaddy webmail; now the dictation box pops up. Not a huge deal, but annoying. There are several other applications that duplicate this process, where in the past, it was just simple dictation to various fields.
My upgrade wouldn't allow me to import my user profiles from 11.5, so I had to start from scratch with a new user profile. More time wasted.
That being said: Dragon is still the leader in speech to text technology (IMO). I'm just surprised that they released this inferior upgrade and called it an "improvement" on previous versions. Purchasers should be well advised to consider an earlier version of the program . . . at least until all of the bugs are worked out of premium 12. Until then, I'm going to dump it an re-install 11.5.
I continue to use version 12, (I have not went back to 11.5, as I stated I would do) and I stand by my above review. Version 12 is very slow to load and often maddingly slow to close. I was hoping that consistent working with the program, including audio training, would improve accuracy. To an extent, it does, but the strange quirks of this "upgrade" just don't make any sense. Often, Dragon takes nearly a full minute to boot up, and when you perform user maintenance to your profile when closing the program, it's not uncommon for this process to take two minutes or more. I might note here that I am using Windows 7 64 bit with 8GB RAM and plenty of space on the hard drive, so there shouldn't be any space/speed issues. I think Dragon rushed this one out, and I hope they make some serious improvements with future upgrades.
I am closing out Dragon 12, and the program is performing its 'User Profile Mainenance.' So far, it has taken over 4 minutes, and the status bar doesn't even read 50%. This is ridiculous, compared with earlier versions of the program.
On another note: those of you using Dragon for their iPhone will be pleasantly surprised that the new update they released just a few days ago does nothing but CRASH! Tried uninstalling and reinstalling the program twice; no luck. Sure, it's free, but Nuance seems to be intentionally driving away customers. There are newly-emerging companies that are competing in this field of limited players, and this loyal customer (until recently) is happy to explore other options. I now see they've lowered the price for Dragon 12 Premium (again), so maybe they figured out that their most recent releases weren't quite ready for prime time. Do yourself a favor and pick up vers. 11.5. I've learned my lesson, and am dumping 12 in favor of an older, better version. And I mean it, this time.
As suspected, 11.5 (the previous version) is an upgrade from their "upgrade." Much faster to load and close, speech recognition is (IMO) better than 12. I can't wait for their customer service folks to call me with a 'special deal' when they release their next upgrade. I will demand a refund for vers. 12 before I purchase their latest "upgrade."
One other note that I would add to those considering purchasing vers. 12: When you are dictating in a text box (as I had to do for my e-mail when using vers. 12), there is a good chance that you will LOSE what you are dictating if you do not transfer your paragraphs piece by piece. Numerous times, I'd compose lengthy emails, transfer them, only to have them vanish. VERY FRUSTRATING AND A WASTE OF OUR TIME, NUANCE. When I went back to vers. 11.5, it allows me to dictate directly to the compose window. Much more efficient.
On a positive note: Dragon recently updated their iPhone app, and (surprise!) this one actually WORKS!
STAY AWAY. STAY COMPLETELY AWAY FROM DRAGON. Although speech-to-text is wonderful technology, my experience with Dragon has been miserable, ever since my 'upgrade' to 12. My most recent venture was an upgrade to a new Dell computer with Windows 10. Dragon crashes within the first few minutes after bootup, every time. On my desktop computer, I play 'Dragon Bingo,' as I don't know what program it's going to decide to be incompatible with. Sometimes it works with Word, sometimes it doesn't. Sometimes it works with WordPerfect, sometimes it doesn't. Sometimes it works with Libre Office, sometimes it doesn't. After all these years, Dragon should just hang it up. The speech to text program that comes with Windows is, at the very least, solid. Before spending the money on Dragon, you should really ask yourself: "Do I enjoy sharp, heavy objects rammed into my ears at high speed?" Because that's what it's going to feel like after you install this program. I might add that their 12.5 'disservice pack' as I call it will cause your hard drive to projectile vomit. STAY AWAY.
on August 27, 2012
I have been testing speech recognition on the PC for some years. Yes, I remember spending ages training software to my voice and still ending up with a system that made constant errors. By contrast, the current Dragon (and to be fair version 11 and 11.5 as well) are superb. Setup takes only a few minutes, and provided you have a good microphone and a quiet environment it makes few mistakes. In fact, I am dictating this review in Dragon 12 and so far it is perfect.
If Dragon does need correction, and there always will be corrections either because of ambiguous words or simply because you make a mistake, the correction box is also very good.
Fundamentally then, this is excellent software and I do not know of anything better for windows. However I did find a few annoyances. First, I could not get good results from the supplied headset. Dragon's own software reported the microphone either as failed or just acceptable. For my test, I went back to an older Plantronics headset that came with an earlier version of Dragon.
Another thing I found annoying is that just occasionally Dragon pauses for a significant time while parsing your speech.
There are a couple of other things for which I do not blame Dragon. One is that it only partially works with Windows 8; that is, it is fine for standard Windows desktop software but does not seem to work in the new "Metro" environment. Second, the Dragon add-in causes Word 2013 preview to crash. That is understandable since Windows 8 is only just finished and Office 2013 is unfinished; but worth noting if either of these is important to you.
Another question is whether it is worth upgrading. My view is that the upgrade from 11 is probably worth it, but from 11.5 hard to justify unless you need one of the new features.
Disclosure: I am a professional writer and was sent this for review.
I've been using Dragon products for some time. Version 11 was the first one I really liked thanks to significantly improve accuracy and not terribly excessive training time. This version, version 12, is tons better. You can get very good accuracy recording through microphone after a few minutes training. It also seems to make even more effective use of contextual information it learns from reading my existing emails and documents to figure out words and phrases I tend to use.
Another plus is that it installed quickly and effortlessly on my 64-bit Windows 7 system. I haven't seen any response time or other performance problems. That may be because I'm using a fairly powerful, quad core, 3.2 GHz Intel I5-based system. Certainly the more capable your computer the better it will perform.
The one thing you still need to get used to is reading punctuation. That does take some getting used to and then becomes second nature. In fact, I sometimes think that slowing down to say "Period", "Comma", and "Double quote" helps me think about what I'm doing.
Version 12 also more effectively transcribes recordings made with a digital tape recorder. After spending a few minutes training Dragon to recognize my voice with an Olympus DS-2500 I'm borrowing, I used it to record this review. The results you see here are exactly how Dragon transcribed that recording.
BTW, I feel that honest, effective reviews can take the place of first-hand experiences that are lacking in online shopping. I've always appreciated the help I've received from other reviewers and work hard to return the favor as best as I can. Please let me know if you found this review helpful and if there was anything you thought was lacking or unclear leave a comment and I'll do what I can to fix it.
on October 18, 2012
First, let me tell you a little bit about myself.
I'm a writer, and I've been using Dragon NaturallySpeaking premium for several years and through several versions. I was fortunate to receive a review copy this time, but previously I have purchased this software, and I consider that money well spent.
I prefer the premium versions because they can read back what I've written, which I find very helpful for editing. Listening to the text can help you pick up things your eyes might miss, like repeated or missing words. I'm currently editing my next novel, and I find Dragon NaturallySpeaking indispensable for that.
At 40 words a minute, I'm also not the fastest typist, so Dragon NaturallySpeaking has been very helpful with that. It's helped me with NaNoWriMo (I had to type that, but maybe I can teach the program to recognize it) two years running. Yes, I've written 50,000-word novels with this thing in under four weeks, and I had a much easier time doing it than I would if I had to type.
I do not hesitate to recommend this program to my writing friends, particularly those dealing with carpal tunnel syndrome or fibromyalgia. I don't know if I could get through NaNoWriMo without it. (Wow, I just trained it to write that, and it got it right away! Amazing.)
I'm dictating this right now, and I am astounded by the accuracy. I've been using Dragon for many years, and this is definitely the most accurate version I've seen -- ever. What's even more amazing is that this is the first time I'm using this version. Okay, it did update my profile from the previous version, but wow! To be this accurate right out of the box just blows my mind.
Aside from increased accuracy, this latest version has other changes that I'm extremely pleased with.
First off, you can now say commands like "file" without the program opening a new file. I hated that in previous versions. Now you have to say "click" for commands to register. That makes a lot more sense. And if you've used a previous versions, you also know the frustration that comes with being told that Dragon can't work in this program or that. Now it opens a text box that you can dictate into. You can then copy and paste the text into PowerPoint or whatever. This latest version is also compatible with more programs, like Internet Explorer 9, Mozilla Firefox 12 and higher, and Google Chrome 16 and higher.
I like the online tutorial, which should be helpful for new users. I did notice one funny thing about it, though. On pages where I was given a list of things to say, the website wouldn't recognize the last item on that list. It did this page after page. Don't worry, it won't reflect your actual experience with the program. I'm currently dictating this into Word, and everything is running smoothly.
As I tell everyone I recommend Dragon NaturallySpeaking to, the one thing I don't like about the package is the included headset. It's rubbish. I've gone through a variety of headsets, and for now I'm most happy with a Plantronics GameCom USB headset, which is comfortable, well-constructed, has great sound, and a great microphone. Yes, I know it's a gaming headset, but it works very well with Dragon. Until a few months ago, I was working with a more expensive Sennheiser USB headset, and while I loved it, it's started to disintegrate. The GameCom headset seems to be made of sturdier stuff. It's also reasonably priced.
So to all my writing friends--whether you're a novelist, a nonfiction writer, a journalist, or a blogger--unless you can type as fast as you can talk, I would highly recommend that you get this version of Dragon NaturallySpeaking with a good headset. If you have a computer that can handle it, I'm confident you'll be happy that you did.
on October 18, 2012
I purchased DragonSpeak for the sole reason that it advertised v12 as Bluetooth compatible. I wear a bluetooth headset many hours a day and find it a comfortable way to communicate. I do not find the wired headsets similarly innocuous. Can't stand them.
After setting up DragonSpeak imagine my dismay when, after 45 minutes on the phone with customer service, I learned there is ONLY ONE commercially available bluetooth headset that is compatible with DragonSpeak. That ONE headset is the Plantronics Calisto. The USB dongle from Plantronics, it turns out, is required, too. A generic bluetooth dongle will not be recognized by DragonSpeak, making this $130 item an undisclosed necessary purchase to bring about the advertised capability of "bluetooth compatible".
Obviously, this is a blatantly deceptive claim.
on October 4, 2012
I have used every version of Dragon's speech recognition software since 5.0 and have seen a steady decline of reliability and even operating system safety of the software over time (see my detailed review of version 11 on Amazon), accompanied by HUGE increases in the sizes of program and speech files, but only modest improvements in accuracy, despite the relentless advertizing of Nuance to the contrary. Many problems with apparent recognition issues in earlier versions simply disappeared with the use of better microphones, and even 11 and 12 will struggle if your microphone does a poor job with the so called 'presence' regions of human speech. All this suggests that the program has become increasingly swollen and bloated in terms of its code base, with the opportunities for bugs expanding correspondingly. If it is possible for a program to be simultaneously the best and worst program running on PCs, this might be it - capable of truly amazing transcription, and at the same time, incredibly dangerous to your system under the right (wrong) conditions.
I upgraded to version 12 hoping to escape the problems with 11 - 11.0 was a disaster, and 11.5 better but still buggy and crash-prone. These included many lockups and even trashing of operating system files and poor interfacing with Office 2003. I was initially concerned when the process of upgrading 11.0 speech files locked up repeatedly - a standard feature of 11.0 when doing housekeeping functions such as "accuracy improvement". I was even more alarmed when shortly after installation of NS 12, Office 2003 became inoperable (and I lost my Outlook .pst file to a hard drive error), and the system blue-screened, and then would not boot due to widespread hard drive errors (a so called 'hard' error where the block is unreadable and cannot be read or repaired without reformatting). I suspect that this happened due to its poor interfacing with an older version of Office (Office 2003 - which itself is way too 'cozy' with the operating system core - such that crashing Office 2003 could trash several system files). But of course I can't prove this, and I sure as shootin' would not repeat the experiment to find out. I also believe that this was due to NS 12, as it was identical to what happened while running 11.0 and 11.5 on two other occasions, so I have to believe that this recent corruption of the OS is NS 12 at work (or at least NS 12 interacting fatally with some other programs - see below).
Of course, this is not ironclad proof that NS 12.0 is the culprit, but please contact me if you have a similar experience (drdougwatt at gmail.com). It is possible that there is some profound incompatibility between MSE (Microsoft Security Essentials), NS 12 and Office 2003 (with perhaps Flash also in the mix somehow?). Had to completely turn off MSE to get NS 11.5 to work well. But again, this is Nuance's failure to really debug their code and make it play nice with other common programs. Flash incompatibility is another possible issue - Abode Flash and Naturally Speaking are both among the most buggy programs running in the PC world and cause more issues than almost any other program and from that standpoint, it's not surprising that there might be some bad interaction between NS 12 and Flash.
Nuance seems to have once again released a really buggy program with very inadequate beta-testing, much like 11.0, which was frankly a mess and totally unreliable. I would caution users to shut the program down when it becomes unresponsive. The previous version (both 11.0 and 11.5) also would not be cooperative with Microsoft Security Essentials which had to be deactivated for the program to work. I know that this is not coming from viruses or other rogue programs. People should be VERY careful about this software. NS 12, along with Flash may among the most dangerous program running in the PC world. I have never seen such widespread damage to a hard drive take place so quickly - from no errors (check every other day with OS disk check) to hundreds of drive errors in a matter of minutes). This is on a new SSD, so no chance this is due to a head crash or other mechanical problem.
I do tech support for my friends on their Windows systems, and usually I can fix almost anything without having to reinstall the OS from scratch. I have had to do three drive reformats from these OS meltdowns (fortunately had multiple backup OS partition images . . so only lost days of work in each instance). My point is simply that I doubt that this was from anything other than a malignant program interaction. Please do email me if you have had a similar experience, esp. if you run NS 11/12, MSE, Flash, and Office 2003. Somehow in the interactions between these four VERY complex programs is the basis for an occasional bluescreen that also results in multiple trashed files and hard drive errors. This only has to happen once in a while for this to be a serious vulnerability. Those without OS partition backup would have to start from scratch, as system restore won't work in these instances - yup - all the way back to a reinstall of the OS, all your programs, and all your files. This is simply too serious an issue for Nuance to do their head-in-the-sand denial (their typical posture around virtually every major issue). That's really the source of the problem - well-entrenched corporate arrogance and lack of real humility about a 4 GB program (speech files plus code base) possibly having liabilities they don't know about. How could virtually anyone be sure about all the functional nooks and crannies of 4GB of code? That's simply too much program to have any real control over interactions and hidden bugs. All of this grief of course obscures the very good to great accuracy that the program can generate, with the right microphone. There is simply no excuse for this kind of behavior in an expensive program. Attempts to talk with Nuance about this set of issues resulted in them reading their advertizing to me. Buyer beware. Nuance doesn't know what they are doing, and they believe their press releases more than user reports. Watch out.
UPDATE October 24th, 2012.
Another complete system meltdown with the same symptoms before the disaster - hard drive was completely trashed and unbootable after being perfect the previous night in a complete diskscan. I was fully prepared with backup OS (operating system) and files. It appears to be a fatal SSD-Dragon interaction, as all these hard drive and system file errors have appeared while these three programs were up and running: prodromal symptoms were that Dragon started to mis-behave while using it to dictate into a browser window that was open and running Flash. I was also running Microsoft Outlook, and in all three instances, I lost the Outlook .pst file. In those instances, I was also using the Crucial M4 SSD.
Using Dragon, while surfing the net or running multiple other programs with known bugs like Flash, is simply unsafe. Nuance has not fixed this vulnerability, which has existed since 11.0. 11.5 didn't fix it and neither does 12.0 - despite their claim to the contrary and their awareness of the issue. Given that Nuance is committed to denial and advertizing over problem-solving and reality-testing, I can't recommend this product, but if you use it, don't use with Firefox, or any other browser that uses the other PC disaster waiting to happen - Flash. If you don't have a full system backup, you'll be very sorry you didn't. There may be other bad actors in the mix to create this very serious OS failure, but it is clear that NS 11/12 is a required element - I have never seen hard drives corrupted like this under Win 7 without Dragon being up and running (and I have owned and operated 8 Win 7 systems). So if you decide to use this program, watch out - and if you have a similar corruption of your system disk, please do contact me, as I am assembling a case report file to send to Nuance.
You have worry when reviews are this completely polarized (lots of one star and five star reviews) - perhaps there are systems on which this program works relatively well, and then others where it really misbehaves - just don't expect that Nuance has any real handle on these issues. They don't - they really don't know what they are doing. Watch out.
UPDATE December 26, 2012
Well since changing how I run the program - never ever allowing the program to go in and out of suspend multiple times and never ever running the program with Flash running in an open browser window simultaneously - I have not seen any fatal hard drive errors emerging on my system. I do think that the program is simply too large to be stable or reliable - it's worth remembering that with speech files the program is significantly larger than the whole operating system itself - suggesting that Nuance might make a more concerted effort to get the bloated code base down significantly. It staggers me that I can load my operating system on a superfast SSD in 12 seconds but it takes 15 seconds to load NaturallySpeaking and my speech files. Something is seriously wrong with those numbers - as no one has ever accused Microsoft of writing svelte operating system code.
UPDATE July 4, 2013
I appear to have solved the problem of the program trashing the file system. I have made several changes in terms of how I use the program, including frequent reinstalls of the program when it starts to malfunction, shutting the program down whenever it starts to misbehave, and rebooting the system from scratch in those circumstances. I also switched to a Samsung SSD after having to reformat my Crucial M4 multiple times. Since making all these changes, I have not had a single hard drive error appear on the system. I'm not sure what's changed - was this vulnerability somehow a function of the previous SSD or drive controller interacting with a buggy program? I doubt that it was the hard drive, as it has been reliable in another system (one without NS 11 or 12 on it). I honestly don't know, and I don't know anyone else who knows a clear answer to these questions either. In any case all this suggests that you should AVOID Crucial M4 SSDs when running this program, and you probably should avoid using Adobe Flash concomitantly. If the program starts to make mistakes with transcription, or locks up or behaves in an unreliable fashion, shut it down completely, and reboot the system.
Independent of how the program might corrupt certain drives, the program does appear to damage itself from time to time when it crashes, along with taking down whatever operational macros it installs into Word/Office. When it misbehaves, it frequently crashes Office 2003, and I have reinstalled Microsoft Office probably 5 times in the last year - presumably again from those operational macros misbehaving somehow. However, despite all this BS, I have not seen any fatal hard drive errors, and it has not trashed my operating system partition in the last year.
UPDATE November 2013
I have been able to duplicate the hard drive error problem, by reinstalling the previously used Crucial M4 solid-state drive, waiting till the program started to interact badly with Microsoft Office and not shutting the program down immediately. I suspect that the issue some how is an interaction between a buggy program, Microsoft Office 2003 (and where evidence suggests that various iterations of Microsoft office are way too cozy with the operating system itself), and some aspect of the hard drive controller software that Crucial/Micron uses to control the SSD. The fact that I have been able to replicate this reliably three time suggests that this is not an artifact, or a one-off variation, but something intrinsic to the interaction between the SSD, and Dragon's code base. Of course there is overwhelming evidence that NaturallySpeaking does not work very well with various versions of MS Office, so perhaps crashing Office brings down some subset of operating system routines . . . but who knows - tracing this out would require professional level software skills - a lot of time, neither of which I have.
In any case, Samsung's hard drives appear to use a different and newer controller and I would recommend anyone using a solid-state drive AVOID CRUCIAL SSDS WITH THIS PROGRAM. Although substitution away to a different hard drive controller does not resolve the intrinsic bugginess of this program, it does seem to eliminate the profoundly serious side effect of occasional drive failure and trashing of the operating system partition. For these reasons I have given the program an upgrade to 2/5 stars. It's still a testimony unfortunately to Nuance's lack of attention to details and customer feedback, and anything other than their own grandiosity. It's a shame, because a debugged version of this program that actually played nice with other common PC programs (particularly Flash and Microsoft office which are both ubiquitous and rather buggy themselves) would be a truly great program. But that would require Nuance to actually listen to customer feedback as opposed to their own advertising and I hold out little hope for that. So that's my final opinion - an amazing program from the standpoint of its potential, but one that is consistently limited by its character-disordered parent corporation that simply won't acknowledge its bugs and unreliability. I believe that this Jekyll and Hyde nature explains why reviews of this program are so seriously polarized.
on November 2, 2013
Before you purchase Dragon 12 you may want to read their privacy policies, and if you're okay with them, then you'll find the program works okay with Windows 8. It is slow though.
I'm sure those who are die hard fans of Dragon will vote this review as "not helpful." That is a common practice with online reviews even when someone is giving an honest experience. However, my review is simply informational. If you're alright with the habits of Nuance then there's no problem for you, buy their software and enjoy it. It does work.
Update: If you go to the Nuance community forum you'll find a lot of people concerned about the forced installation of "Common Software Manager." ( Forum link: [...] )
These types of 3rd party programs can, in some cases, be vectors for malware infecting your computer. The link to remove the Common Software Manager: [...]
on October 16, 2013
To give some context, I have been using Windows speech recognition on Windows 7, the built in voice recognition also known as WSR for probably close to eight months. I recently purchased Dragon premium for work, so I thought I would write a quick review of the pros and cons of WSR versus Dragon premium.
In a nutshell, I would say that Dragon should be your choice if you want to dictate a lot of text. They advertise that it is 99% accurate, and to be frank I believe them. On the other hand, I would say that Windows speech recognition should probably be your choice if you're more interested in voice commands and getting around the Windows interface and less so interested in dictation. Now on to the more detailed points:
I will start with Windows speech recognition, because that is the system I was using first:
- price: it's free, other than getting a reasonable microphone. You really can't beat free.
- It has a pretty nice tutorial that helps you get set up, and helps train WSR on your voice. It's pretty much mandatory that you do this training, otherwise WSR really doesn't seem to do a good job out-of-the-box.
- Show numbers: this is a fantastic command. Essentially it's great as a backup to allow you to click on pretty much anything on the screen. It really helps you navigate around, especially in websites and browsing. One note is that you will have to use this in Internet Explorer, since the implementation in chrome is somewhat lacking. I have not tried Firefox however. (Just one note in chrome, you will get nothing in the webpage itself, unless you enable accessibility by going to chrome://accessibility, but even then I found implementation there lacking).
- "Click this click that click just about anything". I find that in Outlook, I am able to do more with WSR. In particular I can easily go to different folders, by saying click followed by their name. This does not seem to be the case in Dragon. There is a command to go to inbox but not to any specific folders.
- WSR macros. There is a full macro "programming language" that you can get for free from Microsoft that lets you do all sorts of great stuff. You can have shortcuts for typing long things, or for automating just about anything. A few of the shortcuts that I've added are click, double-click, right-click. Out-of-the-box WSR does not allow you to click right where the mouse cursor happens to be at that point in time. ( although out of the box will let you saying right click blah to right click a particular element) But you can add macros do this. I find this very useful since one of the main causes of RSI is holding the mouse and clicking it at the same time.
- Along with the WSR macros, this is very specialized, but I am able to write macros that will actually insert text into a virtual machine VMware running Cent OS. This helps tremendously since I have to type a lot of items into a terminal window. I have not figured out an equivalent way to do this in Dragon.
- You can switch into typing mode by saying "start typing", at which point you can just say the letters or numbers along with modifiers like control shift etc. this is also useful for quickly navigating email and other things.
- dictation could be a whole lot better. Since this is software that came out with Windows 7, you have to remember that that was quite a few years ago. It seems that the core engine has not changed since that time, whereas Dragon naturally speaking has been updated more recently. To be honest, every time I had to reply to an email, I would really think about whether or not it was worth replying, because there would be enough errors to make it a really frustrating experience. I would spend more time correcting the errors than dictating the email the document. On the positive note WSR does have a voice-based correction mechanism, which I only mention, because other tools like Google Chrome's web speech API based recognition does not have something like that.
Onto Dragon premium:
- as I mentioned above, dictation is really Dragon strong suit here. Two things, it's accurate, and it's fast at recognition. Definitely faster than the built-in WSR in Windows. Nowadays when I have to reply to an email of any length I really don't dread doing that at all. It's actually quite rare that I have to correct anything. And when I do most of the time the alternate choice shows up in the correction dialog so it's pretty quick. (As you might imagine, I am writing this review in Dragon NaturallySpeaking 12 premium, not in WSR)
- Dragon does come out-of-the-box with a few things that are nice like saying "mouse click", saying "mouse right click" and "double click", which definitely also help with RSI.
- Dragon has a nice sidebar and also has a few modes like command mode, dictation mode, number mode, spell mode.
- There seems to be somewhat of an eco system built up around dragon. There are 2 extra products that I know of one called voicecomputer and the other called knowbrainer which look like they enhance the product in several ways but they do cost extra money too.
- it's not free :-) I would definitely say that if you do any sort of dictation it's absolutely worth it, especially the home version which is quite inexpensive. I'm not quite sure yet what you get in the premium that you don't get in the home since they both use the same speech engine.
- You can do some amount of command programming in premium, but only for Text-and-Graphics. There is also a macro language for commands but you have to pony up the $500 for the professional version in order to use that. For my use cases, for example, it would be nice to map "mouse click" to something shorter which I have done on the WSR side. You actually click the mouse probably close to about 1000 times a day, and saying "mouse click" that many times is actually quite tiring.
One other mention:
I mentioned Google Chrome's web speech API and recognition. This is a server-based recognition, and I believe it's the same as the one used by the android keyboard. I will mention that you can access this from Google Chrome, there are some extensions that let you do this, and the dictation there is actually reasonably good. But you cannot do things like correct via voice. Also obviously there is no InContext dictation or command capability in that case.
It's kind of odd, it seems that my pros list for WSR is a lot longer than pros for Dragon. But what I would say is this: Dragon is really really good at dictation and in particular the speed and accuracy. Errors take you very long time, and take a lot of energy as well. So even if the accuracy difference is say 85-95% (WSR) to 97-99% (Dragon), each one of those percentages can have a massive multiplier effect. After you dictate you have to do the following steps:
- you need to read what you dictated to check for an error and then identify the error.
- you have to decide what you're going to do about it. This will be one of three choices:
1. Just say "delete that" or "scratch that", and redictate.
2. Say "correct that" and try to correct the entire phrase, Here again is where dragon shines through. Since it tends to be more accurate in the first place, usually the choices it comes up with in the correction dialog tend to be more correct, and I found that I will usually find what I'm looking for in the choices near the top. In WSR I find it to be less often the case.
3. Decide that you just want to correct part of the phrase you just dictated. This means that you will either have to manually select the couple of phrases or couple of words that you want to correct (with your mouse or keyboard), or tell the software that you want to correct that particular phrase by saying "correct start word through end word". Again this is another area where Dragon shines. Too many times with WSR, when I say "correct blah through blah", it will actually dictate that instead of selecting the text I want to correct., which means I need to delete that and try again. With Dragon it tends to get and select what I tell it to.
And in steps 2 and 3 above you will usually have to redictate again.
In any case, you can probably tell that the above steps take significantly longer than the dictation itself, so any percentage points in improved accuracy save you significant time and energy. In fact I would say that every error correction probably takes anywhere from 4 to 12 times the amount of time to correct then the initial dictation, especially with WSR.
I must apologize for being so long-winded, but hopefully somebody will find this useful when trying to decide whether to stay with Windows speech recognition or try out Dragon. You can actually use both, not exactly at the same time, but pretty close. So you can achieve the benefits of both.
Edit December 2:
A few edits after using Dragon NaturallySpeaking for a while longer.
If you are the least bit technical, I would encourage you to take a look at a tool that is free. Just google Natlink, Vocola and Unimacro. Natlink is the lower level component, and Vocola is a very simple command editing language if you will. It isn't the easiest thing in the world to install, but definitely doable. It will allow you to define commands for different applications, and has many samples to work with. Just some examples of things that I have done that may be useful for people so you don't get voice strain:
Instead of having to say mouse click, I've added macros to say go, do, be, click.
For web browsing instead of having to say mouse control click, I've added macros to say "cc". I use this all the time to open new tabs in the background. In Windows speech recognition, you can usually say cc link, where "link" is some text in the link you want to go to. Dragon NaturallySpeaking unfortunately does not seem to be very good in this regard. Usually I have to hover over the link and say cc. When I'm doing heavy web browsing like going through some forums, I tend to use WSR
for scrolling through websites instead of saying page up and page down, I've added macros to say drop N (where N is the number of pages, or if I just say drop, it will only page down once). I use the moniker "sup" to do the equivalent on the page up side. This may sound kind of silly, but when you are browsing many many webpages and forms like I tend to do, saying page up and page down over and over does get very tiresome.
I use Outlook a fair amount, and have added quick steps which you can use to do things like moving to a folder etc. you can map these things to keyboard shortcuts everything from control shift 0 to control shift 9, unfortunately, as you can imagine saying press control shift blah, over and over again is also quite tiresome. I have personally met these to letters, so saying M, will file things to a miscellaneous folder for example. It's as simple as mapping the sound to the key combination shortcut.
In any case, Vocola definitely allows for some sophisticated stuff as well, like moving the mouse to a window edge and dragging it to resize, among other things. But the simple stuff is actually quite easy to do and I would encourage anybody who's the least bit technical to take a look.
For WSR, show numbers still rocks my world, as does web browsing using WSR versus Dragon NaturallySpeaking. As I might have mentioned before there are two tools one by the name of voice computer, and another by the name of show numbers plus, which may help in this regard, but I have not dished out the necessary cash to take a look at those yet.
on September 28, 2012
I am a longtime user of NaturallySpeaking and have really enjoyed the productivity boost of voice-recognition, however after upgrading to version-12 from version-11 I am very disappointed with the product's huge step backwards in performance inside Microsoft Office. You might want to wait to upgrade until they address these issues if you intend to use this product for serious work.
If you edit larger Word documents for example, expect this product to SERIOUSLY impact Word performance and functionality. Once NaturallySpeaking injects itself between you and your keyboard/mouse input and Word, you may find holding down the arrow key to navigate one character at a time in Word drops to a crawl as each character movement takes a second or more and your CPU is pegged. After holding down the arrow key it then takes like 10 seconds for it to catch up and stop moving the cursor. This is of course a 100% unacceptable trade-off between voice-recognition productivity in exchange for significantly degrading your Word interaction experience.
Another example: In a larger Word document in Outline Mode where all Headings cannot fit on the screen, NaturallySpeaking causes an auto-scroll 'feature' to get invoked and you find yourself fighting to keep the insertion point where you want it as the program causes the screen to scroll like you are spinning your mouse wheel. Worse yet, try double-clicking to highlight a word in Word and NaturallySpeaking first scrolls the screen then lets your command through and an entirely different word gets highlighted than the one you selected - Horrors!
One of the reasons I upgraded from Version-11 was that it started having a crashing problem that appeared to be due to a compatibility problem with Flash player. Indeed, search for Flash on Nuance support page and it says to uninstall Flash or roll it back to a previous version. Indeed, uninstalling Flash player reduced the number of crashes NaturallySpeaking was causing. But then your browsing experience is compromised w/o the latest Flash player... When it crashes it crashes ALL your Office applications and you have to go through the Office recovery process for all the documents you had open. NaturallySpeaking also leaves a zombie service process running that you must remember to manually terminate after the crash. Leaving in other applications open that NaturallySpeaking hooked into seems to keep all the zombie terds around so you find yourself rebooting in frustration...
Upgrading to version-12 seems to have solved the Flash compatibility problem (Nuance sales indicated awareness of the Flash problem and stated version-12 resolves it). It almost seems like the "solution" to the Flash crash problem was to intentionally inject pauses or other slow-processing steps to work-around the issues with the way NaturallySpeaking hooks into ALL applications running on your system (i.e. why is it interacting with Flash in the first place when I'm just trying to dictate into Word?).
Submitting technical issues to Nuance support for version-12 finds the typical "what version of the OS are running... how much RAM do you have... what other applications are you running...". I'm likely to return this product within the 30-day guarantee period since it is rapidly approaching. I'll update this review if a more positive outcome materializes. There upgrade special of 1/2 off is ending, so maybe they generated enough revenue to fund a real fix(es) to their compatibility and performance issues of late. In the meantime, their management decision to push this new release out (before testing it?) has frustrated the h*@l out of one of their customers at least...
All that said, I do notice a slight increase in dictation accuracy. If you only work with very small Word documents and are a casual voice-recognition user, perhaps you won't run into these performance issues as much. This product can really boost your productivity.
I'm giving this version 1-star since it is the Premium version but renders Word usage useless due to unacceptable performance impact. If we can get the high-end sports car to lose it's new boat-anchor sparking in the rear-view mirror, I would give this product 5-stars again.