Dictating Numbers

[rickyt29]

Is there a way to make the dictation know that your expecting a number.  So I'll say the 7 for example and dictation makes this SEVEN.  Is there a way to let it know your expecting a number to increase the odds of it getting it right? 

[Pfeil]

Not with dictation, as far as I'm aware.

However, in beta v1.7.10 and up, the "Wait for spoken response" action can be used to listen for specific words or phrases with the same accuracy as command recognition(still keeping in mind the fact that longer phrases are more likely to be recognized consistently).

[rickyt29]

It would be nice if you would say listen for a number and any value it comes up with could be translated to a number.  It's weird from 10 on up it usually makes a number but anything under 10 it makes it a word. 

[Lazza]

 "It's weird from 10 on up it usually makes a number but anything under 10 it makes it a word."
Not really, in English writing number's under ten are usually, or conventionally, written as a word.

[Pfeil]

From what I can see the "Wait For Spoken Response" action actually does return digits.

E.G.

Code: [Select]

Wait for spoken response: '[1..10]'
Will return "1", "2", "3", "4", "5", "6", "7", "8", "9", or "10".

[Gary]

It would be nice if you would say listen for a number and any value it comes up with could be translated to a number.  It's weird from 10 on up it usually makes a number but anything under 10 it makes it a word.

What is it that are you trying to accomplish?  There's different ways to handle different situations.  For example, you can always do replacements in your text if you absolutely need it while dictating.

From what I can see the "Wait For Spoken Response" action actually does return digits.

+1

It does, but that is because it's not using dictation

"It's weird from 10 on up it usually makes a number but anything under 10 it makes it a word."
Not really, in English writing number's under ten are usually, or conventionally, written as a word.

+ 1

[rickyt29]

I'm using it for iRacing, I'm dictating certain laps to do certain things.  So like I say "Set a reminder"? Then I have VA say what lap number? And I start dictating, and I say 5 for example.  And it will return five.  I tried to create a command to replace the dictation text with 5 if it had five, but I'm guessing you can't modify the {DICTATION} using the set command?  Because it's not changing it.  I might have to make another variable and store it in, but it's going to take a lot of work because I built a whole bunch of commands around this method. 

[Pfeil]

Tokens are not variables, so you can't modify their value. Even a "{TXT:}" token, for example, only retrieves the value of a given variable as text, it does not otherwise represent that variable.

If you want to use dictation, you'll have to use a conditional tree, and a variable to store the actual number.

However, if you use the "Wait For Spoken Response" action, which can do what you're using dictation for with better accuracy, the result of that is stored directly into a text variable.

You'll have to make modifications to your profile either way.


For more information on tokens and variables, you can press F1 while VoiceAttack has focus, which will open VoiceAttackHelp.pdf in your preferred PDF reader.

[Gary]

I would totally do what Pfeil is suggesting about the, 'wait for spoken response',  as that will make what you are trying to do way easier and accurate.  99.99% of what is in the beta now is going to full release soon, so, don't feel like you're living dangerously by getting it

[Tazling]

What if it's a hex number?

My numeric dictation wishlist includes numbers like 0003EF2A.

sigh.

[Pfeil]

The number you mention would require over four billion phrases to predefine the possible variations that may be spoken, so that is not an option.

A quick Google search also suggests that the Microsoft Speech Recognition engine does not have a mode/lexicon for recognizing hexadecimal numbers, so dictation would not be feasible either.

That leaves speaking the number digit-by-digit, which as noted can be done using the "Wait For Spoken Response" action (by adding the characters "a" through "f", separated by semicolons, to its list of possible responses).