Asked — Edited

No Answer From The Pc Speaker

Hello,

today i work a little bit with the EZB Software.

I make a new Project and i added 1x "Speech Recognition" and 1x "Speech Settings"

I made a new Phrase

for Example "Hello Robot" and Command Say( "Nice to see you" )

So everything fine .. my Phrase will be recognize ( i see it in the small window) but no sound form the PC speaker ( I must be " Nice to see you")

So the easiest thing, but i dont get it!

What i do wrong?

Boris

PS:

I still work without a EZB4 Controller only Notebook + Webcam/Mic


ARC Pro

Upgrade to ARC Pro

Subscribe to ARC Pro, and your robot will become a canvas for your imagination, limited only by your creativity.

#33  

Hi David!

Hi to all!

I trained yesterday a little bit with the speech reco and cool i must say.

But i have a new qeustion.

Can i make my windows speech reco better if i buy dragon nuance?

Or can i change in the windows settings the speech reco from windows original to dragon nuance?

Because i can train single Words and full sentences in Dragon, for Windows i only can make this Startup trainning.

Boris

#34  

No, ARC uses the Microsoft speech recognition engine.

To use dns, you would have to use the ezrobot sdk to have this take place outside ARC, and pass what was said into ARC via a variable.

#35  

Hi David,

ok i understand, but i can change in the option the speech reco.

Or for what i have this pull down menu?

I remember i had on xp the dragon 10 and i change the speech reco to dragon.

User-inserted image

PS: this pic is from my windows 7 now

And if i can change the option in windows to dragon, also EZB ,ust use this.

I think easylie so.

Boris

#36  

So here is how it works with programming.

You have to specify the speech engine you will use in the code. You can include both in the code and give the user the ability to choose which one they will use, but, the code has to include the engine that it is going to use. This is outside of the windows settings.

ARC uses the windows speech recognition libraries or engine. Even if the OS is set to something else, it really doesn't matter. It will use what is included in the code which is the windows speech recognition libraries.

#37  

HI,

That change the Language to german and i change the voice ( what perfect works with EZB) is something diffrent?

So no chance to change the code ? without the expensice dragon licence

I mean if i speak (mens voice) the reco from windows works 80%

if my girlfriend say the same commands it works only to 50%

I remember i had in Dragon a dictionary where i can try single words with diffrent speakers.

#38  

You could use the SDK to utilize DNS. You would then pass variables back to ARC with what was said, and then script what to do when this variable changes. You would need to buy the DNS developer edition to do this also which is pretty pricy. I have looked at doing this for EZ-AI and just don't have the money lying around to make this purchase.

Unfortunately, Microsofts engine isn't as good as DNS's right now. My hope is that Windows 10 will bring some improvements in this area. Your girlfriend would have to train windows speech engine also if you wanted good results from her speech.

If you are going to use ARC, unless changed at some point in the future, you would be stuck with Microsoft's speech recognition.

#39  

We've been wanting to use DNS for years and this subject has come up often in the past. DNS is so much better then MS's speech engine. Using EZ SDK is an option but 99% of us wouldn't know how to do that. I think DJ said in the past the there are licensing and/or coding issues if that option was added to ARC's background operating code. Now it there were a third part plug-in like Dave's EZ-AI that could do this and have all the coding already done this would be a jump ahead by light years.

Just a thought; I've seen this forum come together and raise money to send kids a EZB. If Dave is confident enough that he could incorporate DNS into his EZ-AI and is willing we could start a funding drive to raise money for him to buy a DNS developer edition. Either that or could there be one or two effluent and interested backer that could fund it them self's? It would have to be understood up front that this is a trial and error emolument and could fail. Also that there is no time line and could take a long time to get either a working or non-working result. How much does a licensed copy cost? What ever, I'm in and ready to pledge.

#40  

The cost for the license that would allow someone to distribute what they have done is somewhere around the $2K dollar mark. I would be willing to put in the time to do this if it were funded. The code for interfacing with ARC is all completed and thoroughly tested on multiple machines.

The thing that you would have to remember is that there would also need to be an application written to train DNS on your computer. I have no idea what this would take but I really don't think it would be hard to do. I would be willing to throw my had in the ring to do the work and program this app if others were willing to jointly fund it.