Really bad recognition

[Jate]

I've looked through all the other forum posts on this, so I apologize to make another post that's so broad, but I'm getting awful recognition right now. I'm using the demo version at the moment, but I will absolutely buy it if I can get past these things. I plan to use this for livestreaming.

Things of note:

- I have it always listening currently, but it doesn't read my voice right away. If I say "forwards", it almost always reads as "wards".

- The recognition is really bad for me right now.
   - I've created a new profile and fully trained it.
   - I tried to download other versions (11 something or other?) but I'm not sure if it worked. I only see 8 something in the recognition panel.

I put Google Assistant next to the fully trained Windows, and it wasn't even close. Google didn't get a single word wrong, and after MUCH fighting and being very staccato with Windows, I finally got it to read one phrase correctly.

I know this can't be right, so any thoughts on what's going on?

Thank you!!

[Jate]

This should have gone in "Issues", sorry about that (if a mod can move it for me).

[Gary]

It certainly is not right, as Windows' speech engine is usually pretty good when used in this context.  What forum posts did you read?  I have to first direct you to these threads as they usually solve just about every speech engine issue:

https://forum.voiceattack.com/smf/index.php?topic=1635.0
and
https://forum.voiceattack.com/smf/index.php?topic=63.0


Normally, things just work straight away, but, if you're using a machine that you've been tinkering with for a while, it could even be something as weird as Google being associated with your headset mic and Windows' speech engine being associated with a webcam (when device selection is that flexible, it can happen).


If you are only seeing speech recognizer v8, you probably have not installed all the files for Speech platform 11, or, you haven't gone into Options > System/Advanced and have chosen, 'Use installed speech platform speech engines' (or both).  VA favors the built-in speech engine and if it detects it, it does not try to switch to the speech platform engine automatically (you have to set it yourself).

Hope some of that helps!