NOMATCH when voice input/end of voice output overlap
Posted: Tue Dec 04, 2007 2:31 pm
I have a VXML application I've recently ported over from BeVocal to Plum. Most things have gone quite smoothly. However, I'm experiencing one general problem throughout my application:
After a .wav file is read to me, it's waiting for voice input. Bargein is generally set to true. If I barge in, my voice is being matched to the grammar correctly. If I wait a second or so (just less) it also recognized my voice response very well. However, if I time things so I respond just as the .wav finishes being played out, I'm getting a lot of NOMATCH events. It seems like the cusp between the sound being played and input being interpreted causes some kind of interpretation quality loss.
I didn't experience this with BeVocal. I don't have any BeVocal system properties set. I'm running with sensitivity set to .60. Are there any Plum system properties I can set to help alleviate this problem?
Thanks -
Matt
After a .wav file is read to me, it's waiting for voice input. Bargein is generally set to true. If I barge in, my voice is being matched to the grammar correctly. If I wait a second or so (just less) it also recognized my voice response very well. However, if I time things so I respond just as the .wav finishes being played out, I'm getting a lot of NOMATCH events. It seems like the cusp between the sound being played and input being interpreted causes some kind of interpretation quality loss.
I didn't experience this with BeVocal. I don't have any BeVocal system properties set. I'm running with sensitivity set to .60. Are there any Plum system properties I can set to help alleviate this problem?
Thanks -
Matt