Asterisk Guru

AsteriskGuru Archives
Mailing List Archives

FAQ

Memberlist

Usergroups

Profile

[asterisk-speech-rec] Speech engine problems

AsteriskGuru Archives Forum Index -> Asterisk-Speech-Rec

View previous topic :: View next topic

Author

Message

vze2fjw3 at verizon.net
Guest

Posted: Mon Jul 02, 2007 2:36 pm Post subject: [asterisk-speech-rec] Speech engine problems

Hi all, I'm having some trouble getting the basic pizza demo working. I have the license server running on one machine, and the SRE and asterisk on another. I have the licensing working, and my dialplan is working now. However, when I try to call, it appears that the results from the SRE are not being returned to asterisk properly. I've included a snippet from the SRE logs to show that it appears to be recognizing "takeout" and "delivery" with reasonable accuracy. The dialplan, however, always returns "0" in priority 4 of the pizza-delivery section, resulting in the execution of priorities 5 and 6, and a loop for input thereafter. I am using an IAX client to connect via FWD. I am using headphones and a builtin microphone for better quality. I originally thought this was the problem, but from the logs it seems that the recognition is okay (though I am a relative newbie, so perhaps there is something there that indicates all is not fine). I am using asterisk version 1.4.1 and 7.1.215 of the Lumenvox SRE. Any suggestions on how to debug or proceed? Thanks, Joe 07/02/2007 13:11:41,608,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] Load grammar time: 0ms 07/02/2007 13:11:41,609,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] Context-Free Grammar Activated: 4 ms 07/02/2007 13:11:41,609,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] LumenVox(161): LM("NO_OOV") deleted 07/02/2007 13:11:41,623,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] LM= "NO_OOV" 07/02/2007 13:11:41,624,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] nbest will not be used on this decode run 07/02/2007 13:11:41,624,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] Begin Core Decode Port 07/02/2007 13:11:41,625,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] starting search 07/02/2007 13:11:41,630,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] uttproc_end_utt 07/02/2007 13:11:41,632,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] uttproc_result 07/02/2007 13:11:41,657,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] winding up utterance... 07/02/2007 13:11:41,657,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] processing final frames... 07/02/2007 13:11:41,658,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] ...last frames processed 07/02/2007 13:11:41,658,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] Deactivating models... 07/02/2007 13:11:41,658,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] deactivating root models 07/02/2007 13:11:41,658,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] deactivating non-root models 07/02/2007 13:11:41,659,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] deactivating word models 07/02/2007 13:11:41,659,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] deactivating singleton models 07/02/2007 13:11:41,659,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] computing lattice density 07/02/2007 13:11:41,660,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] computing phoneme perplexity 07/02/2007 13:11:41,660,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] post processing the forward search 07/02/2007 13:11:41,661,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] preparing acoustic confidence scores 07/02/2007 13:11:41,662,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] forward pass search finished07/02/2007 13:11:41,663,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] begin creating lattice confidence metrics 07/02/2007 13:11:41,665,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] end lattice confidence metric calculation. elapsed time: 2ms 07/02/2007 13:11:41,665,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] word 1: <s>(SIL ) sf: 4 ef: 6 score: 0.123102 ascr:-581632 tscr:0 edge_score: 1.000000 07/02/2007 13:11:41,666,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] word 2: T EY K AW T(T EY K AW T ) sf: 7 ef: 50 score: 0.762551 ascr:-7857152 tscr:-379573 edge_score: 1.000000 07/02/2007 13:11:41,666,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] word 3: </s>(SIL ) sf: 51 ef: 67 score: 0.369225 ascr:-2967552 tscr:-5385 edge_score: 1.000000 07/02/2007 13:11:41,666,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] ...all wound up 07/02/2007 13:11:41,667,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] End Core Decode Port 07/02/2007 13:11:41,669,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] populate answers from hypothesis 07/02/2007 13:11:41,670,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] create answers 07/02/2007 13:11:41,675,0,[AmericanEnglish][MODEL_LOW1:ThreadNdx:0] decode time, Total: 70 ms 07/02/2007 13:11:52,968,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Begin Decode Using Context-Free Grammar 07/02/2007 13:11:52,970,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Loading Context Free Gramamr 07/02/2007 13:11:52,971,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Retrieving words 07/02/2007 13:11:52,972,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Word retrieval time: 0 07/02/2007 13:11:52,972,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Load grammar time: 0ms 07/02/2007 13:11:52,972,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Context-Free Grammar Activated: 4 ms 07/02/2007 13:11:52,973,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] LumenVox(161): LM("NO_OOV") deleted 07/02/2007 13:11:52,987,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] LM= "NO_OOV" 07/02/2007 13:11:52,988,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] nbest will not be used on this decode run 07/02/2007 13:11:52,988,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Begin Core Decode Port 07/02/2007 13:11:52,988,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] starting search 07/02/2007 13:11:52,993,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] uttproc_end_utt 07/02/2007 13:11:52,995,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] uttproc_result 07/02/2007 13:11:53,025,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] winding up utterance... 07/02/2007 13:11:53,025,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] processing final frames... 07/02/2007 13:11:53,025,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] ...last frames processed 07/02/2007 13:11:53,026,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] Deactivating models... 07/02/2007 13:11:53,026,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] deactivating root models 07/02/2007 13:11:53,026,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] deactivating non-root models 07/02/2007 13:11:53,026,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] deactivating word models 07/02/2007 13:11:53,027,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] deactivating singleton models 07/02/2007 13:11:53,027,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] computing lattice density 07/02/2007 13:11:53,027,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] computing phoneme perplexity 07/02/2007 13:11:53,028,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] post processing the forward search 07/02/2007 13:11:53,028,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] preparing acoustic confidence scores 07/02/2007 13:11:53,031,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] forward pass search finished 07/02/2007 13:11:53,031,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] begin creating lattice confidence metrics 07/02/2007 13:11:53,033,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] end lattice confidence metric calculation. elapsed time: 1ms 07/02/2007 13:11:53,033,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] word 1: <s>(SIL ) sf: 4 ef: 7 score: 0.141685 ascr:-865280 tscr:0 edge_score: 1.000000 07/02/2007 13:11:53,034,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] word 2: D AX L IH V AXR IY(D AX L IH V AXR IY ) sf: 8 ef: 68 score: 0.875996 ascr:-10625024 tscr:-379573 edge_score: 1.000000 07/02/2007 13:11:53,034,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] word 3: </s>(SIL ) sf: 69 ef: 77 score: 0.116462 ascr:-1543168 tscr:-5385 edge_score: 1.000000 07/02/2007 13:11:53,034,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] ...all wound up 07/02/2007 13:11:53,035,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] End Core Decode Port 07/02/2007 13:11:53,036,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] populate answers from hypothesis 07/02/2007 13:11:53,037,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] create answers 07/02/2007 13:11:53,042,0,[AmericanEnglish][MODEL_LOW2:ThreadNdx:0] decode time, Total: 80 ms _______________________________________________ --Bandwidth and Colocation Provided by http://www.api-digital.com-- asterisk-speech-rec mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-speech-rec

jcolp at digium.com
Guest

Posted: Mon Jul 02, 2007 2:50 pm Post subject: [asterisk-speech-rec] Speech engine problems

Joe Hosteny wrote:

Quote:

Hi all,

I'm having some trouble getting the basic pizza demo working. I have
the license server running on one machine, and the SRE and asterisk on
another. I have the licensing working, and my dialplan is working now.

However, when I try to call, it appears that the results from the SRE
are not being returned to asterisk properly. I've included a snippet
from the SRE logs to show that it appears to be recognizing "takeout"
and "delivery" with reasonable accuracy. The dialplan, however, always
returns "0" in priority 4 of the pizza-delivery section, resulting in
the execution of priorities 5 and 6, and a loop for input thereafter.

I am using an IAX client to connect via FWD. I am using headphones and
a builtin microphone for better quality. I originally thought this was
the problem, but from the logs it seems that the recognition is okay
(though I am a relative newbie, so perhaps there is something there
that indicates all is not fine).

I am using asterisk version 1.4.1 and 7.1.215 of the Lumenvox SRE.

I would highly suggest you update both your Asterisk and engine. If you
are using a relatively recent connector build of 7.1 (even the last one
done) then you are probably running into a version skew scenario as the
API changed.

--
Joshua Colp
Software Developer
Digium, Inc. - "The Genuine Asterisk Experience" (TM)

_______________________________________________
--Bandwidth and Colocation Provided by http://www.api-digital.com--

asterisk-speech-rec mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-speech-rec

Display posts from previous:

AsteriskGuru Archives Forum Index -> Asterisk-Speech-Rec	All times are GMT
Page 1 of 1

You can post new topics in this forum
You can reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum