Other Sites
One of the catalysts that persuaded me to finally make the switch to Intel from my faithful Power PC laptops was that MacSpeech's Dictate voice recognition software, which is based on the superb Dragon NaturallySpeaking voice engine, does not support Power PC.
I'm a moderately heavy user of dictation software. My travails with fibromyalgia and presumably related neuritis symptoms over the past decade plus have made long typing sessions difficult, especially when symptoms are flaring. Some keyboards are better than others, and many are just plain impossible because using them quickly causes symptoms to flare if they're not already rampant (I don't have RSI per se, but keyboarding does aggravate things), and I rely a great deal on dictation software, without which I would have long since been out of business as a professional writer.
Consequently, while my own interest in dictation software, at least at first, stemmed from my chronic health issues, dictation programs actually address a variety of users along a spectrum from "need" to want",. The most urgent sort of "need" is for people with major disabilities who simply cannot type. Then there are folks like me for whom typing is possible, but less comfortable, increasingly so when done for long periods of time, and for whom speech recognition makes computer operation and composing prose digitally much more comfortable. And, as the MacSpeech folks observe, given enough hours of straight typing, just about anyone can fit in that category. Then there are those who are motivated by the potential for increased productivity - writers who find oral composition flows naturally, and just anyone who discovers that entering words into electronic documents by voice can be substantially faster than even a very proficient typist can achieve working on a keyboard. Consequently, MacSpeech Dictate is designed to accommodate users in all of those categories. I would want to have it around now even if my health issues were somehow cured.
The first (I think) dictation application for the Mac was Articulate Systems Power Secretary, which was absorbed by Dragon Systems in the mid-'90s, then sold to the British developer GT Interactive Europe, which continued to market it for a time as Voice Power Pro 3. Power Secretary and Voice Power Pro used discrete dictation engines, meaning that they needed you to pause momentarily between each word uttered , which was an unnatural mode of expression most found fatiguing after even short stints. That said, it was better than no dictation software when your arms were aching and your wrists and fingers numb and burning.
The first continuous speech dictation product for the Mac was IBM's ViaVoice Millennium Edition, which I eagerly switched to in the late '90s. Unfortunately, continuous speech proved to be vastly more demanding of hardware power than discrete speech, and ViaVoice ME ran sluggishly to say the least on my then workhorse Mac - a 233 MHz G3 series PowerBook, and while it was less tiring to use, I didn't really achieve much, if any, advantage in dictation speed over Voice Power Pro on that hardware. The 233 MHz G3 processor just couldn't keep up, even when I spoke very slowly.
Another deficiency of ViaVoice ME was that except for a handful of specified applications, none of which I used, you had to dictate into a dedicated window called SpeakPad - then transfer your text via the Clipboard to its intended destination. You couldn't dictate into Finder text fields, spreadsheet cells, and so forth, or control Finder functions with your voice.
Incorporated in 1997, exclusively Mac-oriented MacSpeech first registered on my radar screen in October, 1998, and the company's first preview of its dictation product for the Mac was announced in December, 1999, distributed on CD at MacWorld Expo 2000 New York in July of that year. Their iListen 1.0.1 continuous speech dictation software was finally released just in time for Christmas 2000, with MacSpeech commendably not fudging the fact that it was still very much a work in progress, and offering a free update to iListen 1.1 for early adopters (it arrived in November, 2001). However, even in that unfinished, essentially late beta build, iListen offered functionality never before available on the Macintosh - continuous speech recognition that could dictate into virtually any application or text field. On the downside, I found that iListen's FreeSpeech 2000 Speech Recognition engine from Philips Speech Processing was not as efficient as the speech engine IBM used in ViaVoice, at least in recognizing my voice. I found iListen needed a lot more training than ViaVoice did. I could get decent recognition performance out of ViaVoice after reading it just one story, better with two, and I never bothered reading it more than three. I could read six or more stories to iListen, and it still appeared to want more training, especially for single word and short phrase accuracy. Happily, iListen did prove to be a tolerable performer on my not exactly cutting edge Mac hardware, and I actually achieved tolerable performance with it on a 200 MHz 604e SuperMac S-900. The superior speed and the sense of freedom you got from being able to dictate into virtually any program or text field was liberating after being cooped up in ViaVoice's SpeakPad.
I switched pretty much exclusively to iListen as it progressed through many upgrades over the balance of my Mac OS Classic years, and it was progressively better with each revision.
However, IBM beat MacSpeech to market with an OS X supporting dictation application in 2001 with ViaVoice X, which was fantastically better than the old Millennium Edition of VV and better than the contemporaneous version of iListen as well. It was fast, even on a 500 MHz G3 Pismo PowerBook, delivered unprecedented recognition accuracy with minimal training, and supported dictation directly into most applications and text fields. I only read two of the six training stories and didn't bother to analyze my documents. Very impressive. On the downside, I found the program somewhat cranky and experienced difficulty getting it to install that involved a conference call with IBM tech support. That issue was successfully ironed out, however, and for a time I switched to ViaVoice X as my dictation app., but not for long. For one thing, IBM quickly lost interest in developing ViaVoice for the Mac and farmed the program out to ScanSoft in 2003, at which point development essentially ceased.
In the meantime, MacSpeech got their first OS X native version of iListen out the door in August, 2002, with iListen 1.5, which I was delighted to find had improved or eliminated many of the issues I'd found annoying and frustrating with iListen 1.01 through 1.2.1 for the classic Mac OS, although I still found its voice correction capabilities lame. I also still rated ViaVoice X better for speed and accuracy, but I essentially switched back to iListen with version 1.5 anyway and never went back to using ViaVoice for production.
iListen dictation really became a mature product with the release of version 1.6 in 2003 - the first OS X - only version of the product. Mac OS Classic users were still accommodated by an upgraded version 1.5.5 incorporating some but not all of the version 1.6 enhancements came bundled with iListen 1.6,
Refinement of iListen continued through several revisions os 1.6, version 1.7 (the first Universal Binary version of the program), and version 1.8, which was the last build of iListen before MacSpeech terminated the program's development to concentrate solely on Intel-only Dictate, which was introduced at Macworld Expo 2008, where it received the 'Macworld 2008 Best of Show' award.
MacSpeech Dictate 1.3 - The Review
So here we are in April, 2009, 39 months after Steve Jobs rolled out the first Intel Macs at MacWorld Expo 2006, and I at last have a MacIntel — a lovely little jewel of a 2.0 GHz unibody MacBook with a Core 2 Duo processor that is frankly overkill for most of the stuff that I do with computers, but whose processor muscle is tailor-made for running dictation software. Speech recognition software used for dictation is more complex than most other types of consumer software, and with arguably more "moving parts" then in just about any other kind of application software.
Now, I have great affection for iListen has a familiar and much appreciated tool, but I have to say off the top that Dictate is a much better program in every way that I can think of. Dictate is an entirely different animal from iListen, although there are some familial user interface conventions and commonalities that make the transition from the older program to the new mostly smooth and intuitive.
The most apt descriptive for Dictate that springs to mind is "effortless." With iListen, even at its best in version 1.8 on the G4 laptops, one always felt like there was work and effort involved using it. You had to remain conscious of sloppy pronunciation and so forth, Which can be a challenge late at night when you're already propping your eyelids open with toothpicks trying to stay lucid to meet a deadline. With Dictate, I'm finding that I can just babble away, and still find accuracy impressive with only the initial five minutes or so of training. The Dragon NaturallySpeaking speech engine is quite obviously superior to the Philips FreeSpeech 2000 Speech Recognition engine used in iListen.
Dictate comes with two Installer disks, a Plantronics microphone headset and necessary USB adapter, and a quick start guide pamphlet. The very comprehensive user manual and other documentation are supplied on disk in PDF format.
Installation is reasonably straightforward, although it does involve inserting the two installation disks in sequence, and the data files one takes quite a while to export its contents to your hard drive. You're also obliged to register the software at installation time, which requires an active Internet connection, leading me to wonder what one is supposed to do if they don't have access to an Internet connection.
Once the installation process is complete, it will be necessary to train the program to recognize your voice efficiently. However, this step is not nearly as time-consuming or onerous as it once was. I'm getting just amazing accuracy after only reading the five-minute initial training story, a treatise on speech recognition. Transcription speed is close to real time even with the modest 2 MB of RAM I have in this machine - much better than the previous high water mark of ViaVoice X in its day.
MacSpeech recommends using voice for commands, correction, and navigation as much as possible, and vigorously discourages using a hybrid mode or partly dictation and partly keyboarding/mousing, maintaining that alternating between straight speech recognition and keyboard/mouse is "a fantastic way to confuse the software." I have to confess that in many years of using iListen, I've found both navigation and text correction slower and more cumbersome than just interpolating with the keyboard and mouse or trackpad. I actually have both a hand and a (clicking only) foot mouse connected at my main workstation, and use them in tandem both to spread mousing stress and to speed things up, which using two pointing devices simultaneously can do.
However, I've been trying to break myself of the hybrid entry habit, and I perceive that Dictate's very lively responsiveness is going to help greatly with that effort.
The Plantronics mic that comes bundled with Dictate seems to be excellent, at least as good as the Parrott mic's that I've been using with iListen 1.8. The headset is light, adjustable, attractive looking, and it has a comfortably long cord. The mic does require that USB dongle, which is supplied, but it's small and evidently works just fine although it does use up another precious USB port (I have two 4-port hubs connected to my MacBook and find that no more than enough). However I was using a Griffin iMic USB adapter with the Parrot mic. for iListen, so no change there, and both of these microphones are approved for use with Dictate.
When dictating to your computer, it's necessary to speak the punctuation, which I've never found terribly difficult to get onto, although on occasion I have absentmindedly found myself, particularly when reading aloud, lapsing into articulating punctuation when speaking to fellow human beings.
Dictate has four modes of speech recognition: Dictation Mode, Command Mode, Spelling Mode, and Sleep Mode all switchable among using your voice alone. Dictation Mode is self explanatory but while you're in it you can also issue commands, In Command Mode the program recognizes and executes only commands - thus can be looked on as "Command/Only Mode."
Spelling mode was completely revamped with MacSpeech Dictate 1.2, allowing users to spell words, names, or acronyms by using either natural language or the International Radio Alphabet. Sleep Mode simply lets you toggle the microphone and application on and off.
Another new feature introduced with Dictate 1.2 was Phrase Training, which helps you increase accuracy by letting you train your voice profile as you proceed. MacSpeech notes that most of the speech recognition industry refers to this feature as "correction" which is accurate in the sense that it does can be used to correct text in your document, but contends that this somewhat misses the point because ongoing training also helps refine MacSpeech Dictate's ability to recognize what you intended to say. With phrase training you are not stuck with the level of accuracy from your initial voice training until you take the time to read more training stories, and you don't have to specifically train it from a document or selection. Instead you can train a phrase immediately when you see it was misrecognized. Then the next time you dictate that phrase or word in the phrase, MacSpeech Dictate is more likely to recognize it accurately.
The way you proceed with this is when you notice that the program's transcription has misinterpreted a word, for example substituting the word "cause" when you intended to write "pause," while still in dictation mode you can say "select the word cause." Confirm that the word has been correctly selected, then say "pause" and check to ensure that the word actually has been changed to "pause." That's all there is to it.
Spelling mode and phrase training work much the same way. You pause between navigation or spelling/correction instructions and general dictation in order to allow the program to distinguish between the two.
With phrase training, you speak the phrase you want to train, then open the recognition window which will display several possibilities from among which you can select the correct phrase using its respective button, or say "edit one" "spelling mode" [ pause ] and spell out your words. MacSpeech notes that sometimes it can take two or more attempts before Dictate learns to properly recognize a particular phrase.
Yet another addition to Dictate 1.2 was the "new move command" which facilitates more sophisticated navigation within a document man was supported by the initial versions and iListen before them.
For someone with a physical handicap involving a disability that renders even small amounts of typing difficult, the Command and Spelling modes would be a priceless help.
However, for just about any user, Dictate's voice transcription capability is a dream come true, and it's so accurate that not a whole lot of correction is necessary anyway, although at the level of training I've given it so far, it does usually require some. I also find it useful for instant messaging or composing short emails or whatever if the program is already up and running. It's seductively slick.
That said, Dictate is not perfect. I wish there was some way to hide the floater control palette (which is alot prettier than the old one in iListen) when it's not in use other than closing the program, and I'm still not a fan of dedicated notepads, a la the not fondly remembered one from ViaVoice Millennium Edition,
However, Dictate's Notepad application, according to MacSpeech, will always provide the most optimized results, particularly when executing advanced features, complicated navigation, and massive amounts of editing with your voice, although I've found performance quite satisfactory and my favorite application for composition and editing, Tex Edit Plus.
The program still takes a relatively long time to start up, which I suppose is somewhat inevitable when it has to load a voice profile and dictionaries before you can get underway. Adding more RAM, which I plan to do in the not too distant future, or a 7200 rpm hard drive, which I don't, would probably speed things up some more.
I wish there were a way to download version updates as standalone installers using a browser. The Dictate 1.3 update for example was 12,240k, which translates to about two hours downloading time over my poky dialup connection, and it can't be paused. For downloads that large, I usually end up having to pause the process several times while I attend to other online stuff. However, the update download and install went smoothly and successfully aside from the time involved.
The latest version 1.3 of Dictate will now work with any document, including ones that were not created with Dictate. Dictate 1.3's new "Cache Document" command that allows MacSpeech Dictate to read a document's contents, in order to navigate the document, and to enter and edit text.
Also introduced with MacSpeech Dictate 1.3 is an extensive new Help Book system, two new visual microphone status indicators, and new commands for "Cache Selection," "Press The Key," "Press The Key Combo" and "Cancel Training."
The new microphone status indicators provide a quick visual cue of whether the microphone is "on" or "off" by quickly looking at either the new Menu Bar status menulet, or the new microphone status badge on the MacSpeech Dictate Dock icon. To augment the User Manual and other helpful resources, a new searchable, indexed Help Book for MacSpeech Dictate is now available from the Help menu.
MacSpeech Dictate 1.3 also fixes reported issues, adds several enhancements and is free of charge to all registered customers, using the "Check for Updates" feature I mentioned above.
To sum up, in my 2007 review of iListen 1.7 I commented "iListen 1.7 rocks!" Well, Dictate 1.3 rocks even more and is as I said above, in a whole different dimension from iListen. If you've never tried dictation software, or perhaps did some years ago and found yourself underwhelmed, it's time to consider (or reconsider) how it can make your computing more comfortable, pleasant and efficient.
MacSpeech Dictate has the following system requirements:
• An Intel based Macintosh computer
• Mac OS X 10.4.11 "Tiger" or later
• Mac OS X 10.5.2 "Leopard" or later
• 512MB of RAM (1GB or more for better performance)
• 2.0GB of available hard drive space
MacSpeech Dictate sells for $199 (including the Plantronics headset), and is available from the MacSpeech store at:
http://www.macspeech.com
[Disclosure: MacSpeech is a longtime and appreciated Applelinks advertiser, but this review was not solicited nor its evaluations influenced by them, and impressions and opinions on products reviewed on Applelinks in general are the author's own.]
Charles W. Moore
Tags: Hot Topics ď Reviews ď Writing/Publishing ď
Sounds like a great program for handicap people.