R
Roedy Green
I am experimenting with Dragon Naturally Speaking, a text to speech
device.
One of the frustrating but amusing things is that it only understands
the word "backspace" when I sound annoyed. It is training ME to snarl
at it.
It does much better with long phrases. It has a devil of a time with
individual letters.
We will see how it evolves after some training. I have seen demos
where the thing made not a single error even when we in the audience
made up the text.
Specialized speech to text might be very useful in programming since
the vocabulary is limited to the defined variables, methods and
keywords. In theory it should be much easier than general text
recognition, if you could train it on the definitions easily.
In contrast with typing, the longer the variable names are the more
accurate it becomes.
device.
One of the frustrating but amusing things is that it only understands
the word "backspace" when I sound annoyed. It is training ME to snarl
at it.
It does much better with long phrases. It has a devil of a time with
individual letters.
We will see how it evolves after some training. I have seen demos
where the thing made not a single error even when we in the audience
made up the text.
Specialized speech to text might be very useful in programming since
the vocabulary is limited to the defined variables, methods and
keywords. In theory it should be much easier than general text
recognition, if you could train it on the definitions easily.
In contrast with typing, the longer the variable names are the more
accurate it becomes.