Recording: Status-based delete and other questions

[gschaller]

Hello,
I have set recording to ALLCALLS in my campaign (no auto dialing). But then I have also a recording if the lead does not answer, only ringing tone and an agent suspiring. Any way to delete recordings based on status given by agent at the end of a call? So for example delete if status is "NA - No answer".
Next question: For every call I have two files, *-in.wav and *-out.wav. The *-out.wav file has always a size of 44 byte. The *-in.wav file contains the whole conversation. So why do we need mixing of the files? In the *-in.wav file I have the whole conversation (agent and lead is speaking).
My last question is about a way to move the recorded files to a different server (for mixing or converting into MP3). So I can have a storge server which does also the mixing, the Asterisk has one less thing to do. Can I change the extensions.conf this way?
exten => 8309,1,Set(MONITOR_EXEC=/usr/local/sbin/move_files.sh)
exten => 8309,2,Answer
exten => 8309,3,Monitor(wav,${CALLERIDNAME},m)
exten => 8309,4,Wait,3600
exten => 8309,5,Hangup

Thanks for the great support!

[mflorell]

I wrote a script to do something close to that. It's the VDAD_file_clean.pl script in the sphinxNA folder in one of the last couple releases.

It doesn't do exactly what you want, but you could use it as a framework because it does delete recording based upon vicidial status in the database.

[gschaller]

Thanks, I will have a look on this file. Is there a way to change vicidial in that way it begins recording when the lead picks up the phone? And how do you do recording? I tested recording with 10 agents, but had bad sound quality and echo. With 5-6 agents it is ok, so the bottleneck is the hard drive (SATA drive in my server). Now I think about recording in ram-disk.

[mflorell]

In VICIDIAL the recording does begin when the call gets to the agent. This is done because the channel can go through many different routes depending on how your system is setup and as a channel changes what it it connected to you can have issues with recordings.

We typically use regular ULAW or GSM codec recording depending on the application, and we can get upto 50 recordings going on a single server before we run into audio issues. That is with T1s and ULAW only with a low VICIDIAL dial level.

[gschaller]

I have Auto dial level = 0, and there is recording before the lead/customer answers (Zap channels, Digium E1 card). The server is a Dell PowerEdge SC1420 SATA with Xeon processor and 2 GB memory, so should be enough power for recording. I will do some more tests tomorrow to reach some more simultaneous recordings.