It Doesn’t Help to Cuss at Speech Recognition Software − 5 June, 2008
But it doesn’t hurt either. I’ve been fighting some battles with speech recognition software. There are the battles with the Mac OS X built-in speech recognition, and then there are the battles with MacSpeech Dictate. Let’s begin with the MacOS X built-in speech recognition.
I wrote about my dialogue with “Dude” in my article Editing Photo Collections Via Voice Recognition. In that story my dialogue with Dude was fairly civil. Since then Dude and I have had some tussles.
I was having trouble getting the speech recognition commands for MacSpeech Dictate to work with Safari. So I tried defining keyboard shortcuts using Mac OS X’s built in speech recognition. Now, it really hurts me to write anything negative about the Mac. I’ve been a long time Mac fan. But my dialogue with Dude when using Safari sounds something like this:
“Dude, next page.”
(Nothing)
“Dude, next page.”
(Nothing)
“Dude, next page.”
(Nothing)
“Shit.” Long pause. “Dude, next page.”
(Nothing)
“Dude, next page.”
“Whit!” says Dude, moving to the next screen.
“Dude, last page.”
(Nothing)
“Dude, last page.”
(Nothing)
“Dammit.” Longer pause. “Dude, last page.”
(Nothing)
“Dude, last page.”
(Nothing)
“Dude, last page.”
“Whit!” says Dude, moving to the previous screen.
“Dude, last page.”
“Whit!” says Dude, moving to the previous screen.
“Dude, next page.”
“Whit!” says Dude, moving to the next screen.
Now, in order to capitalize “dude” if I have forgotten to do so already, I must select “dude” and say, “Cap dude”. After a conversation like this, I really feel like “capping” Dude. Perhaps the problem is that the built-in speech recognition doesn’t seem to have a way to train on my voice. I know that voice training is very important in speech recognition. Strangely, it’s hard for me to tell the difference between one “Dude, next page” and the next “Dude, next page”. Yet clearly Dude can tell the difference.
Okay, let’s move on to my difficulties with MacSpeech Dictate. Although I have trained the software extensively with writing samples and voice training, it still has trouble recognizing some very basic words. Even after I feed the software training on the words that it missed, it continues to miss some of the same words, forcing me to type.
Worse yet, trying to use MacSpeech Dictate with Apple Mail has disastrous consequences. My only positive experiences using the software with Apple Mail involves dictating the messages. Most of the other things I have tried to do with Apple Mail and Dictate have caused Dictate to hang. Shortly thereafter Apple Mail also hangs, although sometimes it recovers, but not really.
After I submitted about 10 automatic bug reports, including increasingly frustrated messages, I received e-mail from tech support. The e-mail asked some very basic diagnostic questions. Some of the questions were indirect suggestions, such as restarting, reducing the number of applications open at the same time, and creating a new voice profile. Today, I made sure that I had answers for every one of the questions in the tech support e-mail. I am sad to report that not one of the measures that I tried based on the e-mail’s suggestions made one iota of difference.
My final complaint about MacSpeech Dictate is that the two most useful commands for Safari are broken. I tried, too many times to count, saying “move to next link” and “move to previous link” with no result. One might wonder whether I had forgotten about Einstein and the notion that trying the same thing over and over is unlikely to yield different results. However, because of my experience with Dude, I had begun to think that if I said the commands enough times, it would eventually work. But “move to next link” and its brother “move to previous link” never worked, no matter how many times I invoked them.
While reading documentation about how to define commands, I noticed a little “compile” button next to the commands that consisted of AppleScript. I pressed the compile button for “move to next link” and got an error message. The AppleScript was busted. Ditto for “move to previous link”. So I filed another bug report.
I hope that at some point the kinks in MacSpeech Dictate get worked out, because it seems much more promising than the Mac OS X built-in speech recognition. I really want to like Dude, but he’s actually kind of a disagreeable fellow, considering that he’s there to help.
I wrote about my dialogue with “Dude” in my article Editing Photo Collections Via Voice Recognition. In that story my dialogue with Dude was fairly civil. Since then Dude and I have had some tussles.
I was having trouble getting the speech recognition commands for MacSpeech Dictate to work with Safari. So I tried defining keyboard shortcuts using Mac OS X’s built in speech recognition. Now, it really hurts me to write anything negative about the Mac. I’ve been a long time Mac fan. But my dialogue with Dude when using Safari sounds something like this:
“Dude, next page.”
(Nothing)
“Dude, next page.”
(Nothing)
“Dude, next page.”
(Nothing)
“Shit.” Long pause. “Dude, next page.”
(Nothing)
“Dude, next page.”
“Whit!” says Dude, moving to the next screen.
“Dude, last page.”
(Nothing)
“Dude, last page.”
(Nothing)
“Dammit.” Longer pause. “Dude, last page.”
(Nothing)
“Dude, last page.”
(Nothing)
“Dude, last page.”
“Whit!” says Dude, moving to the previous screen.
“Dude, last page.”
“Whit!” says Dude, moving to the previous screen.
“Dude, next page.”
“Whit!” says Dude, moving to the next screen.
Now, in order to capitalize “dude” if I have forgotten to do so already, I must select “dude” and say, “Cap dude”. After a conversation like this, I really feel like “capping” Dude. Perhaps the problem is that the built-in speech recognition doesn’t seem to have a way to train on my voice. I know that voice training is very important in speech recognition. Strangely, it’s hard for me to tell the difference between one “Dude, next page” and the next “Dude, next page”. Yet clearly Dude can tell the difference.
Okay, let’s move on to my difficulties with MacSpeech Dictate. Although I have trained the software extensively with writing samples and voice training, it still has trouble recognizing some very basic words. Even after I feed the software training on the words that it missed, it continues to miss some of the same words, forcing me to type.
Worse yet, trying to use MacSpeech Dictate with Apple Mail has disastrous consequences. My only positive experiences using the software with Apple Mail involves dictating the messages. Most of the other things I have tried to do with Apple Mail and Dictate have caused Dictate to hang. Shortly thereafter Apple Mail also hangs, although sometimes it recovers, but not really.
After I submitted about 10 automatic bug reports, including increasingly frustrated messages, I received e-mail from tech support. The e-mail asked some very basic diagnostic questions. Some of the questions were indirect suggestions, such as restarting, reducing the number of applications open at the same time, and creating a new voice profile. Today, I made sure that I had answers for every one of the questions in the tech support e-mail. I am sad to report that not one of the measures that I tried based on the e-mail’s suggestions made one iota of difference.
My final complaint about MacSpeech Dictate is that the two most useful commands for Safari are broken. I tried, too many times to count, saying “move to next link” and “move to previous link” with no result. One might wonder whether I had forgotten about Einstein and the notion that trying the same thing over and over is unlikely to yield different results. However, because of my experience with Dude, I had begun to think that if I said the commands enough times, it would eventually work. But “move to next link” and its brother “move to previous link” never worked, no matter how many times I invoked them.
While reading documentation about how to define commands, I noticed a little “compile” button next to the commands that consisted of AppleScript. I pressed the compile button for “move to next link” and got an error message. The AppleScript was busted. Ditto for “move to previous link”. So I filed another bug report.
I hope that at some point the kinks in MacSpeech Dictate get worked out, because it seems much more promising than the Mac OS X built-in speech recognition. I really want to like Dude, but he’s actually kind of a disagreeable fellow, considering that he’s there to help.
