Jump to content
Sign in to follow this  
alpha117

Speech Recognition

Recommended Posts

Hi All.

for P2ATC to recognize a phrase, the words spoken must be heard and understood.  This means all the usual microphone and settings issues are not a problem and, because we're using a set of command grammars, the phrases spoken must be in one of the grammars that is active at the time the words are spoken.

This is an area where, if you are saying something and it is not being recognized, I can tweak the grammar options to try and pick up on the words or you can rephrase your speech to match the existing Grammar.  This is where the Grammar Help screen comes in handy.

 

As has been discussed in other threads, the rate of speech can make a difference in whether or not the recognition engine can understand the words spoken before it compares to the active Grammars.  So clarity of speech is also critical.

 

That said, if you think you are saying the right words according to the Grammar Helper, but are not getting expected results, please let me know exactly what you said and take a screenshot of P2ATC or the Grammar Practice screen if you are practicing.  If you're flying, let me know which controller you are connected to and what phase of the flight you are in.  A copy-paste of the text in the speech text window would also be useful. 

Send it to me at the admin email address and I can then do some testing and see if there is something in the Grammars that can be changed.  I appreciate everyone's PIREPs and getting some in the Grammar area will be helpful.

 

Thanks,
Dave

Share this post


Link to post
Share on other sites

In windows 8.1 its speech recognition popup has an entry for:

"Train your computer to better understand you"

which I presume also trains you to speak more clearly for these apps.

I did the session a couple of times (15 minutes each) and it did improve my articulations and/or the computer's understanding of my vowels etc..

 

Would this training infer its computer's improvements directly to Pilot2ATC's recognition software  since they use the same engine?

 

Also, with a text document open and the cursor at the start, you can say start listening, then you can start a speech-to-text session.

There are quite a few words, like vector (it typed in specter) that I had to improve my enunciation of, but am improving in that way. And like I said before, I hope it lends itself to  improvement within P2ATC

 

Once I said : Beech 197 request vectors to final

and it wrote: Beach 197 request  baxter's two final

tried again with excessive accuracy (so I thought ! ) and got this:

Beach won 97 request actors to a final    :search:

and finally after a little self training and separation of words I got:

Beach won 97 request vectors two final  

:Four Leaf Clover:

 

P2ATC "practice" writes in nothing if it gets a wrong or unrecognizable string ... so I think...

 

I would think that P2ATC has its own vocabulary of recognition so it may very well take the input baxter and consider it  vector?

 

How does it work?

 

Suggestion: include a group of similarly sounding  words to be accepted (and even a user supplied list of acceptable pronunciations?)

Share this post


Link to post
Share on other sites

John,

Thanks for the question and training session PIREP. :smile:

To the extent that the training session trains you to enunciate better, as in your example of vector, it will likely improve your chances of recognition in Pilot2ATC. 

On the computer side, this training only trains the computer for open dictation interaction, of which there is none in Pilot2ATC.

 

P2ATC's Practice sessions are listening for any recognized phrases.  If it hears none, it prints nothing on the screen.  It will, on occasion,  "bend" the words to fit a known phrase, so actor might be interpreted as vector if said in a string of words that otherwise fits.  The complexity of listening for all similar words would be overwhelming. 

Training is the best approach here.  It is surprising to me how well it recognizes variations in speech.  The trick is to say the right words  in the right order.  This takes some getting used to, but comes naturally after a while.  The grammar can also be tweaked.  If, for example, the grammar has a phrase like "Beech 1VB is on final" that it recognizes, but you consistently say "Beech 1VB on final" and it is not recognized, just let me know and I will tweak the grammar to not require the "is".  While I've tried to capture all these types of variants, I inevitably have missed at least a few of them.

 

Hope this helps,

Dave

Share this post


Link to post
Share on other sites

Dave,

Thanks for your explanation, will continue to work with it.

 

Here is a little minor glitch for ya...When Anna tells me I am off course, her text prints out,

Beech one victor bravo you are off course. Turn (and then the numbers).  Well you see the period right after the word course? So she says;" Beech One Victor Bravo you are off course dot" kinda funny when I first heard it, thought she was calling me Dot!

 

Chuck


Intel(R) Core(TM) i7-12700KF CPU @ 4.90GHz  32.0 GB RAM Windows 11 Home NVIDIA RTX 3080Ti 12GB Multi monitors

WD SN850 M2 2TB PSU 1000W. 360 AIO.cooling.

 

Share this post


Link to post
Share on other sites

Hi

Great tip in one of replies from Dave was, to switch the frequency back to standby then back again to main, if im not getting reply from atc center etc, this gets me working again, no problems so if you get no reply after speaking to atc try the above

Mhazy

Share this post


Link to post
Share on other sites

Ok Dave,

She did it again but on a different function and I was in the sweaty palm part of a set up for ILS approach and did not note where it was, no big deal. I will see it again one day.

 

Regarding voice recognition.

This time on my flight I took explicit note of the requirements for requests for Approach and followed to the letter,

went perfect, take off to landing. KATW - KMSP.

So, I think we are about to get this all figured out with help from all you  folks.

 

Thanks,

Chuck


Intel(R) Core(TM) i7-12700KF CPU @ 4.90GHz  32.0 GB RAM Windows 11 Home NVIDIA RTX 3080Ti 12GB Multi monitors

WD SN850 M2 2TB PSU 1000W. 360 AIO.cooling.

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

  • Tom Allensworth,
    Founder of AVSIM Online


  • Flight Simulation's Premier Resource!

    AVSIM is a free service to the flight simulation community. AVSIM is staffed completely by volunteers and all funds donated to AVSIM go directly back to supporting the community. Your donation here helps to pay our bandwidth costs, emergency funding, and other general costs that crop up from time to time. Thank you for your support!

    Click here for more information and to see all donations year to date.
×
×
  • Create New...