Showing 6 results.
Items per Page 50
of 1
First Previous Next Last

CVP Forum

« Back to General Discussion

Error on recognition using Nuance VWS2.0 in native audio mod

Combination View Flat View Tree View
Threads [ Previous | Next ]
Error on recognition using Nuance VWS2.0 in native audio mod
Answer
4/10/03 8:33 AM
I can run HelloWorld and any app that just plays prompts. As soon as I speak, a bad fetch vxml exception is generated (according to the audium logs)
10.0.1.150.1049940837968.13.yesno,04/10/2003 14:14:15,AUDIUM SERVER ERROR: A VoiceXML exception occurred: error.badfetch

VWS doesnâ¿¿t seem to generate any call logs when using native audio, so I canâ¿¿t see nuances version of things, but presumably the audium server is generating some vxml that references a URL that Nuance canâ¿¿t find.
Audium server gets this exception, and generates an illegal vxml1.0 page which VWS spits the dummy at, and generates itâ¿¿s own error message. (a known Audium issue).
Where do I look next to find out what is going on? How do I find out what URI nuance couldnâ¿¿t fetch? Is there a nice way to use a browser to capture the vxml that is being served?
My test app is simply a yes/no voice element, but the same thing happens with credit card when I tried that.
Flag

RE: Error on recognition using Nuance VWS2.0 in native audio mod
Answer
4/11/03 4:19 AM as a reply to Anonymous.
Hello,

Can you let me know the following:

  • Does the application always crash on the first input regardless of the element type?
  • Chapter1_HelloWorld does not crash if there is no bargein input?

Thanks,

Joe
Flag

RE: Error on recognition using Nuance VWS2.0 in native audio mod
Answer
4/11/03 6:14 AM as a reply to Anonymous.
yes that's right, input triggers it regardless of element (I only tried Credit card and yes/no though)
Chpater1_HelloWorld just has an audio element so it works.
Flag

RE: Error on recognition using Nuance VWS2.0 in native audio mod
Answer
4/11/03 6:47 AM as a reply to Anonymous.
here are some more clues.
I tested using using AgeIdentification (from the tutorial)
First time I set it to DTMF, and went right trough -no problem, except of course in native audio mode, I couldn't enter anything and after 3 no-inputs it hung up.

Then I changed it to voice only.
I said "twenty" and heard the first no_match prompt back, so it did detect (and reject) my input.
Then I said 18 and got the bad fetch error.
See audium logs attached.
So voice input doesn't immediately and necessarily cause the bad fetch error.
Flag

RE: Error on recognition using Nuance VWS2.0 in native audio mod
Answer
4/11/03 10:16 AM as a reply to Anonymous.
some more clues
It's a problem with nuance VWS not being able to find the builtin grammars. This is a problem configuring VWS, rather than an audium thing, but if you know how to fix it, that would be great. I suspect it has to do with locales.
The recserver log fragment is attached
Flag

RE: Error on recognition using Nuance VWS2.0 in native audio mod
Answer
4/11/03 7:12 PM as a reply to Anonymous.
Hello,

You are correct. This is a VWS 2.0 configuration issue in how builtin grammars are being accessed. We have seen before when your reseller also ran into the same issue. I believe the workaround was to serve the builtin grammars off of a separate server. However, you should contact Syntropy for more details on how the fix was implemented.

Regards,

Joe
Flag