I’ve been fascinated by speech recognition for a very long time. I used a program called Simon on a NeXT computer back in 1992. I have toyed with every version of Dragon Naturally Speaking since v2 (now owned by Nuance). I keep upgrading my copy of Dragon Naturally Speaking (through v9, I haven’t done v10 yet), even though I never actually use it for anything real beyond checking out how much better each version has gotten.
The primary reason I don’t dictate more is that Lois and I work two feet apart 99% of the time. That makes it awkward to be speaking to the computer, from a number of perspectives. Still, I remain intrigued by the concept.
Vista has built-in speech recognition. My new laptop also has Realtek HD Audio built in, including a very high quality microphone next to the webcam, at the top of the monitor. As an example, I tested it with Skype the other day, with no headset, and there was no echo on either side of the conversation, and the other person said I sounded fine.
That made me think about the extra convenience of being able to dictate without scrounging around for headset, or wearing it for extended periods. I decided to play with the speech recognition just to see.
Basically, it works pretty well. Far from perfect. In fact, I’m not sure that Dragon 10 wouldn’t be better. That said, it’s built in, and feels much lighter weight (starts up instantly, shuts down instantly, doesn’t shift application windows around to put its toolbar up, etc.).
It took me a while to get it to work with my USB headset. Basically, you don’t tell the speech recognition program which device to use. In order to use it with the USB headset, you have to set the USB headset to be the default microphone on the system, and then the speech recognition program automatically picks it up.
You might be asking why I wanted to use the USB headset? The simple reason is that my headset has a microphone mute button on the cord. That’s very cool for speech recognition. If the phone rings, or Lois wants to talk, I can just hit the mute button, and speech recognition is off, even though the program is still listening. It simply can’t hear anything. As a bonus, in theory, the recognition should be better, but for now, I don’t care too much about that.
Here’s one annoyance. It was my intention to dictate this entire post, including all of the actual production of it (clicking the save button, publish, etc.). Unfortunately, I gave up after five minutes. I tend to write my posts in Firefox, right in the admin interface of WordPress. Even with Allow Dictation Everywhere set on in speech recognition, it doesn’t think that Firefox is a normal input program (though it recognizes that I’m in a text area).
So, every phrase gets put up in a dialog box for me to confirm. It got them all correct, but I couldn’t just speak the post. I could have dictated into Word, WordPad, NotePad, WindowsLiveWriter, etc., and in the future I might just do that, but for now, I’m typing this post…
Using speech recognition in Command Mode works reasonably well. I can switch applications easily, select menu items, switch folders in email (Thunderbird, which obviously isn’t written by Microsoft), etc. Yesterday, while eating lunch with both hands, I had an IM conversation with someone by speaking my responses and saying Enter after each one. The concept was very cool, even though I had to correct a bunch of words (I wasn’t using the USB headset at the time).
Anyway, I recommend playing around with speech recognition in Vista if you work in a room alone, or have a spouse (or co-worker) who would be amused by your ranting at the computing out loud. 😉