100% call recording

I have created a custom SIP recording server for one of my customers that was tested up to 48 concurrent calls. The testing limit was from the host system not the SIP recorder. We only had two T1's to test with. Based on system load, two more calls should be possible:) The SIP recorder read the sip header to get specialized information for storage and trans coding of the recorded files to selectable audio formats. So what you speak of is entirely possible. It may not be something you would want to do in a single box depending what else you want the trixbox to support, but doable.

Thanks for these - I'll investigate both options.

Below is some information on how I have recording setup on our Trixbox CE system for a 40-50 agent call center.

We currently have around 40-50 agents on our system doing 100% call recording. We have been doing it for the past year and a half using Trixbox. I setup the system to use a ram drive. What the system does is is record the calls to a ram drive and then archive them off every 2 minutes to our storage area.

You need to modify your /etc/grub.conf file to increase the ram disk size because the default will be to small. Below is a snapshot of part of my /etc/grub.conf file. You may need to google some better steps on this because I am writing this part from memory. the ramdisk_size option below is set for around 2 GB (overkill most likely but it works well for me).
title CentOS (2.6.9-42.0.10.ELsmp)
root (hd0,0)
kernel /vmlinuz-2.6.9-42.0.10.ELsmp ro root=LABEL=/ ramdisk_size=2097152 acpi=off noapic
initrd /initrd-2.6.9-42.0.10.ELsmp.img

The you have to do something to load the changes to grub, I forget what the exact command is.

Then you need to setup your system to /etc/rc.local to format and mount the ram drive at boot. Below is a snapshot of my /etc/rc.local. The * sign denotes lines that need to be added for my call recording setup. Note remove the * from the live setup.

#!/bin/sh
#
# This script will be executed *after* all the other init scripts.
# You can put your own initialization stuff in here if you don't
# want to do the full Sys V style init stuff.

touch /var/lock/subsys/local
* /sbin/mke2fs -q -b 1024 -m 0 /dev/ram0
* /bin/mount /dev/ram0 /var/spool/asterisk/monitor/
/etc/trixbox/runonce
/usr/local/sbin/motd.sh > /etc/motd
/usr/sbin/fxotune -s
/usr/sbin/amportal start
* rm /call_mv_running
* rm /convert_mp3

Finally you need a script to copy the files to mass storage. Below is a script of mine that works (I am not the best script writer ever, but it does the job). This script will move call recordings over 2 minutes old to a storage share. In this case it is a /sdb/callrec for us. This is called from a cron job.

#!/bin/bash
i=/call_mv_running
echo $i
if [ -f $i ]
then
echo $i
echo move_rec.sh already running, call aborted
else
echo "move_rec.sh running" > $i
echo no move_rec.sh processes running starting move_rec.sh
find /var/spool/asterisk/monitor/*.wav -mmin +2 | xargs -i mv '{}' /sdb/callrec/
rm $i
fi

I then have a compression script that runs and compresses all the files to mp3. There are 2 parts to it. I am using an ls and grepping for the wav files because my store had to many entries at one time for a find command. These 2 scripts run at night to compress all the previous days calls.

convert_to_mp3.sh
#!/bin/bash
i=/convert_mp3
echo $i
if [ -f $i ]
then
echo $i
echo convert_to_mp3.sh already running, call aborted
else
echo "convert_to_mp3.sh running" > $i
echo no convert_to_mp3 processes running starting convert_to_mp3.sh
cd /sdb/callrec
rm /callrec.txt
ls > /callrec.txt
fgrep "wav" /callrec.txt | while read file
do
/encode.sh "/sdb/callrec/$file"
done
rm $i
fi

encode.sh
#!/bin/bash
export TERM=vt100
echo $1
WAVFILE="$1"
OUT=${WAVFILE%.*}
LAME=/usr/bin/lame
echo $LAME --silent -V7 -B24 --tt $OUT.wav --add-id3v2 $OUT.wav $OUT.mp3
$LAME --silent -V7 -B24 --tt $OUT.wav --add-id3v2 $OUT.wav $OUT.mp3
test -r $OUT.mp3 && rm $WAVFILE

I then have a custom application that uses an archive copy of the cdr db from the Trixbox server and locates the recordings by matching up the unique ids.

An embodiment of the invention is a method of tracking calls from technicians equipped with technician assistance devices. The method includes a call center personnel accessing a call center terminal and the call center personnel receiving a call from a technician. The call center personnel enters call data through a user interface including a plurality of fields, one of the fields being a designation of whether the call from the technician was necessary or unnecessary. The call data is stored. Upon receiving a request from a user, a report is generated identifying the call from the technician as necessary or unnecessary. A system and computer program for implementing the method are disclosed.
---------------------------------------
Jennysmith

Influencer