Author Topic: VA recognition vs Windows recognition  (Read 1004 times)

nzteddy2020

  • Newbie
  • *
  • Posts: 2
VA recognition vs Windows recognition
« on: November 04, 2022, 01:06:18 AM »
Apologies if this has been covered elsewhere.
I am getting very poor recognition in VA.
I have trained up Windows Speech Recognition 3 times.
When I use windows dictate to test common commands I expect to use in DCS, I get virtually 100% success.
However the exact same speech in VA is very poorly recognised.
It was my understanding that VA uses windows speech recognition as the base engine, so I cannot understand why there is such a huge disparity?
Any clues much appreciated.

I have attached what I dictated into notepad. Take my word for it that it went in from speech, I didn't cheat :-)
Also attached a sample log output of me repeating the same thing into VA. The results vary, but it pretty much never succeeds.
All of what I am saying is configured as commands.

I must have something fundamentally wrong?

Pfeil

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4759
  • RTFM
Re: VA recognition vs Windows recognition
« Reply #1 on: November 04, 2022, 05:30:20 AM »
When you mention you're dictating using Windows Speech Recognition, are you referring to the application shown in the attached picture?

Gary

  • Administrator
  • Hero Member
  • *****
  • Posts: 2827
Re: VA recognition vs Windows recognition
« Reply #2 on: November 04, 2022, 09:45:10 AM »
What happens when the profile is loaded is that the prefix and suffix commands you have are joined together to make full (composite) commands that VoiceAttack is listening out for.  That means you have to say the complete commands all at once, without pause.  From the log, it looks like you're saying the portions of the commands separately.   For instance, it looks like you're saying, 'chief', waiting, and then saying, 'turn off the ground power'.  The pause after speaking is making VoiceAttack look for a command called, 'chief' (which is not recognized) and then a command called, 'turn off the ground power' (also not recognized).

What VoiceAttack is really listening for in that regard is a complete sentence of 'chief turn off the ground power' in order to execute the command (one fluid sentence, no pausing).  Try that and let us know how it goes.

nzteddy2020

  • Newbie
  • *
  • Posts: 2
Re: VA recognition vs Windows recognition
« Reply #3 on: November 04, 2022, 11:33:43 AM »
@Pfeil Yes that's the windows app I mean by dictation.

@Gary Ah yes, I think you are onto it! One fluid speech is much better. I might stop using prefixes/suffixes. Talking with the pause is a hangover of trying to get Vaicom going, which I couldn't. So I decided to revert to VA only. Hard to kick the habit of the micropause.

Thanks very much.