Since Scott is writing his blog posts using speech recognition, I decided it was time for me to try this out. I even bought myself a new USB head phone with a microphone. I must admit I'm having a lot of difficulty with it. The first time I was with my son, but he did a lot of noise. This time I'm all alone,
<end of dictation> <good old keyboard>
Boy, I'm not ready for speech recognition. Dictating the above paragraph was a complete nightmare. First, it seems Windows Speech Recognition is having a hard time with my accent. The first time I ran the tutorial, it had a lot of trouble with the word "PERIOD". Even after running the training, it could not insert a period.
I ran the tutorial a second time this morning, and it went very well. But when I returned to Windows Live Writer, I could not undo nor select anything. Saying "undo" or "undo that" displayed "undoing last action" in the speech recognition monitor, but actually undid nothing. Trying to select or correct something was useless, the selection would not change. I even got a bug, where the "Select a number" popup showed, I dictated the number and said "OK" and my change was made, but the popup stayed visible. Pressing or saying "Cancel" or "OK" or even "Close that" did nothing, so I had to click on the close button.
I suspect it's a problem with Windows Live Writer. Let's try writing the next paragraph in Notepad, then paste it here.
<end of keyboard> <start of dictation>
I even tried speech recognition with slimKEYS. I told windows to press the window K keyboard shortcut, and the slimKEYS window appeared. Then I tried to select hotkeys by their names. It went pretty well. I could select the Win letters combinations, but had difficulty with the Browserstop Hotkey.
<end of dictation> <good old keyboard>
I'm gonna cry... It was awful, terrible, excruciating, so painful I feel like jumping out of the window. Sure, corrections and selections worked this time, but I could have typed all this in a fraction of the time and energy I took for the above paragraph.
My conclusion? As long as speech recognition will require the mere mortal to train his computer (and himself) for weeks before getting things done, it will remain a geek gadget and a lost of time.