With that in mind I was probably a bit too enthusiastic, a bit too forgiving in my early assessment of MacSpeech Dictate. Or rather in my assessment of my own abilities to use it straight out of the box. This is an unusual product in that it really asks a lot of the user. If you're not willing to do the training, if you're not willing to understand the parameters of the software, and most importantly if you're not willing to read the fucking manual then forget it. It's not a cheap bit of kit and you'll do your dough cold. If you are willing to invest the time learning how to use MacSpeech (or its Dragon-based Windows equivalents) and training the software to work with you then it could be a really powerful tool.
First question, does it work?
Yes, and it's awesome. It is freakishly accurate, much more accurate than my own typing. And there is no need, as I mentioned in a previous post, to speak in an American accent, Shatner style. (And yes it recognizes the word 'Shatner').
But it's not Star Trek. It won't do all these things out of the box. You do need to train it to listen to your voice, and the software needs to train you to speak to it. The more time you spend using MacSpeech, or Dragon, the more accurate it will become. But only if you take the time to update your profile.
Your profile is the software's understanding of how you speak, and to a lesser extent how you write. Think of it as a super-avatar. It also includes a lot of environmental information, so you might record one profile for dictating in a quiet room and a different profile for a noisy room. When you first start dictating, the program will make errors and just as importantly you will make errors because you'll probably try to speak to it like a person, not just a bunch of code. If you mumble, cough, slur your speech, whatever, it'll all be transcribed.
The first couple of days you'll spend a lot of time pointing and clicking in what's called the Recognition Window. In this little box you'll find the programs interpretation of what you just said, and up to 10 alternatives. Option number one is always Dictate's best guess but if option number six was the right choice you just click on that and it swaps out the copy. If none of the options were accurate you can choose one to edit and use that. After a couple of hours of doing this and of saving your profile as you go, you'll notice a marked improvement in the program's ability to understand you. As you get more confident and you relax you'll also find you're able to speak much more quickly and conversationally until you do get to the point of that Star Trek moment where you just speak at the computer and the words magically appear on screen.
That's dictation, it's not editing. And editing is way more important than composition in terms of whether your finished product is readable. I had real fears that MacSpeech Dictate would be crap at editing, and in one sense it is. If I simply opened up a huge manuscript and tried to edit the thing via voice command I would fail. There is just no way that telling a cursor where to go and what to do is anywhere near as efficient as using a mouse and keyboard. But as all of the reviews and a lot of the documentation that comes with MacSpeech makes clear, you're in for a terrible hiding if you try to mix keyboard and voice commands. It just won't work, and you'll crash the program. As I did at least half a dozen times during one very frustrating hour this week.
Its all down to the cache. MacSpeech/Dragon keeps two things in its mind; what you said and what it wrote. If you fuck up that delicate balance by using your mouse and keyboard instead of your voice you'll blow the cache apart and overwhelm the program.
That could be a deal killer, because of the unwieldy nature of using voice command to edit. There is however a caveat. The software comes with its own notepad, a very very basic text editor in which you can compose your documents. Again most reviews and the software's documentation emphasize the ability to use MacSpeech with most of the other bits of software on your computer, with MS Word, with Firefox, whatever. But here is JB's tip... Don't Do It.
Yes, MacSpeech Dictate can work with all of these programs, but it probably won't. It will almost certainly crash within the first few minutes.
The notepad on the other hand is an exceptional environment into which to dictate. It is stable and robust and most important of all it allows you to edit with your mouse. You can just place the cursor wherever you want, define whatever text you want, and dictate right over the top of it. The essence of editing.
Unfortunately for me, I only figured that out after a couple of very frustrating days of constantly crashing the system. It got to the point where I was so pissed off, so depressed and so fucking desperate that I went back and did what I should have done in the first place. I Read The Fucking Manual from start to fucking finish. In doing that I learned of at least half a dozen very basic errors I'd been making over and over again, and I picked up a whole bunch of obscure but powerful pointers for getting the most out of the program. If I hadn't done that I reckon I'd have thrown it away and there'd have been tears before bedtime.
Bottom line, it works and it can work brilliantly, but whether it does is down to you.
There is one final point I'd make though. It feels weird. I am so used to 'thinking with the tips of my fingers' that, as lobes pointed out earlier this week, I just wasn't writing like normal when I used the dictation program. I'm still not, but I am getting past the initial strangeness where I constantly felt as though I was thinking about thinking about dictating the writing. I suspect that will take a bit longer to get used to than simply mastering the mechanics.
Anyway, my apologies. I am not a software reviewer and this entry has been a useful exercise in teaching me that if I ever wanted to be I'd have a lot to learn. There's so much more I could tell you. Some good, such as its ability to inhale vast slabs of your writing for syntactical analysis which then gets fed into MacSpeech's memory, allowing it to better understand how you write. And some bad, like its tendency to 'hear' your breathing as dictation (mostly a problem when you're sitting, staring at the screen, saying nothing).
So perhaps I should just throw the floor open to questions.