Jump to content


< Back to Forum


 

Voiceguide Crashing When Installed On Azure Iaas


  • Please log in to reply

#1 Accel 11 June 2016 - 12:29 AM

We are trying to set up a Voiceguide IVR on a Windows Server 2012 R2 VM in Azure IaaS. We have connected an Asterisk device to it, and the registration between the two seems to be fine.  However, when a call is placed, which hits the Asterisk device, then the server, a tts file that is supposed to be created does not do so correctly, and the VoiceGuide service crashes.

 

It is running a Intel Xeon E5-2673 @ 2.40 GHz with 28.0 GB RAM.  We have installed VoiceGuide version 7.5, and Dialogic HMP drivers, newest version.

 

I am attaching zip files containing VoiceGuide logs files, and the Dialogic logs. 

 

I'm new to Voiceguide and these configurations, so you may need to speak slowly : )

 

Thank you/

 

 



#2 SupportTeam 11 June 2016 - 08:31 AM

Traces suggest that something went wrong during the TTS generation.

 

and vgTts trace shows:

 

NOTE  TTS engine used is set to first engine in list, as no TTS engine set in VG.INI [SAPI] TTSEngine entry

 

Please edit VG.INI to set the TTS engine to use. The currently used first engine in list is a Microsoft 'Desktop' engine - which may be where the problem is given that this is system is running on Azure (and therefore has no access to a sound card/device ?)

Traces show that a TTS for use with VoiceGuide was installed on this system so please first try setting VG.INI to use that engine instead and see if that resolves the issue.

 

Please see: http://www.voiceguid.../config_tts.htm for instructions on how to set the TTS to use.

 

 

Also:

 

Are you able to remove all TTS use from the script ?

 

Looks like the TTS generated in this case was a static sound file with no parameters, so it can just be replaced by a pre-recorded / pre-generated .WAV

Can other TTS use be replaced by static files as well? Are you able perhaps to use concatenated pre-recorded sound files instead of TTS is other cases?

 

Other attached traces show that playing of sound files was done with no errors and the call completed fine.

 

In general, we we recommend that static files do not get generated using TTS. It is faster for system to use a pre-recorded sound file, so in cases where a pre-recorded sound file can be used - it should be used. The TTS generated WAV files are stored in VoiceGuide's \temp\ subdirectory, so for static playbacks you can just take the TTS generated sound files from there, move it to different directory and amend the script module accordingly to use that static .WAV file.

Also, many TTS can be replaced by concatenating pre-recorded .WAV files that list various options. That approach can be used to give better results then TTS, and we recommend looking into doing this in cases where that approach can be applied.

 

 

Dialogic HMP on Microsoft Azure Cloud :

 

Please also note that Dialogic HMP is only officially supported by Dialogic when installed on own physical servers, or under VmWare ESXi.

Microsoft Azure Cloud is not officially supported by Dialogic - but looks like Dialogic HMP work fine on it, and hence Azure can be used as a platform for VoiceGuide deployments.



#3 Accel 14 June 2016 - 12:48 AM

Thank you for the response.  We specified the correct TTS engine, and it solved THAT problem.  TTS now seems to be working fine.  Since that issue seems to be corrected, it has now raised another issue:

 

Our script prompts a caller to input information with the phone keypad (DTMF).  However, DTMF is not being recognized by the system and there is no indication of it in the monitor.

 

Please advise of what I need to send you for analysis.

 

Thank you.

 



#4 SupportTeam 14 June 2016 - 07:33 AM

OK, Thanks for letting us known that selecting the correct TTS engine resolved the crashing issue.

 

Regarding the new issue can you please start a new topic thread  for this new issue.

 

Please collect a WireShark trace capturing the SIP and RTP for the call, and advise how are the calls sent into system (who is providing the SIP service?) and how are DTMF tones configured to be sent over that service.

 

Please post that information and the WireShark and VoiceGuide traces for the system startup and the call and we can then see what happened.

 

To limit WireShark capture to just SIP and RTP please specify this in the Wireshark's 'Filter' text box:

 

sip || rtp

 

then use File->Export Specified Packets to save the displayed packets into a .pcpapng file.

 

The WireShark packet capture will show if the DTMF signalling is being sent to this system over the SIP/RTP connection.

 

Please .ZIP up all traces before posting.